International Journal of Biometrics (36 papers in press)
Applications of Deep Learning Algorithms in Biomedical Signal Processing Pros and Cons
by Gyana Ranjan Patra, Saumendra Kumar Mohapatra, Mihir Narayan Mohanty
Abstract: The artificial neural networks have the capability of learning any mathematical function with training data. They have found many applications in regression, classification, and other complex problem domains. The term deep learning used in neural networks implies that multiple hidden layers are introduced and is the corresponding network is termed as the Deep Neural Network system. These complex systems can facilitate computational models to acquire data representation with multiple levels of abstraction. The deep neural networks have shown consistently good results in the area of speech recognition, visual object recognition, object detection, and many other domains such as drug discovery and genomics. In this paper, the study on different biomedical signals has been presented. Further, the work is carried out with cardiac signal as mostly cardiac failure reported a lot, and research on this signal occurs a special position in recent days. For verification, authors have approached the ECG classification with a deep learning model. A detailed study of the pros and cons has been accomplished. Also, a methodology for ECG classification using a 10-layer convolutional neural network has been proposed. Results have shown that the proposed model is capable of performing better than some of the other methods like support vector machine and wavelet transformation methods.
Keywords: ECG; ECG classification; deep learning; deep neural networks.
Intelligent Products Recommendation System based on Machine Learning Algorithm combined with Visual Features Extraction
by Jianzhong Yang, Huirong Chen, Xianyang Li
Abstract: When the current methods were used to design the visual recommendation system with product images, the desired product cannot be recommended for the users, recommendation need to take a long time, and there are problems of low user satisfaction and low recommendation efficiency. A method of intelligent recommendation system for products, taking Nixing Tao products as an example, was proposed based on gradient boosting decision tree algorithm in machine learning and Zernike features extraction of machine vision. The overall structure of the recommendation system for products was formed by user shopping Module, System Management Module, Database module and visualization module. The key features of products that were extracted by Zernike Moment, and they were analysed by clustering in order to obtain the association rules between products and users. We embedded the association rules into recommendation model based on gradient extension decision tree algorithm. The experimental results show our method has short recommendation time and good user satisfaction.
Keywords: Machine Learning; Nixing Pottery Products; Intelligent Recommendation; Zernike features; Machine Vision.
Facial Landmark Detection and Geometric Feature based Emotion Recognition
by P. Shanthi, S. Nickolas
Abstract: Facial expression related to machine intelligence is a popular research area in emotion science, pain assessment, human behavior analysis, virtual reality, etc. This paper aims at exploring a contour-based shape analysis from the viewpoint of geometric characteristics towards facial expression recognition. Since the facial landmark detection accuracy dramatically affects the final classification, a simple contour detection algorithm is used for identifying facial landmarks accurately. Spatial local and relative geometric features extracted with the neutral face as the reference are projected to the lower-dimensional space using stepwise linear discriminant analysis. The proposed system is tested and validated using back propagation based artificial neural network on JAFFE and MMI dataset with an average accuracy of 95.53% and 94.98%, respectively. The proposed scheme's recognition accuracy has been compared with the state-of-art methods, and the results show significant improvement in the proposed model over others using geometric features alone.
Keywords: emotion; facial expression; geometric features; stepwise linear discriminant analysis; back propagation based artificial neural network.
Research on the Path of Fitness Strategic Planning Based on \"Internet+\"
by Deliang Wei
Abstract: In the early 1980s, Europe and the United States, the leading economic areas, began to set off a wind of \"national fitness path engineering\". The phrase is mainly derived from the participation of all people, despite the race or group. The places where the national fitness path is built are basically in some relatively beautiful environment, such as parks, lawns, rivers, and so on. With each cycle, a class of sports equipment is installed, and each type of equipment is connected by a path. Its purpose is to exercise and make people healthier, and this is the most important in the development of national fitness project.This study is based on the \"Internet plus\" context, with residents and fitness managers as the research object. In the course of the research, the author mainly combines the literature method, interview method, questionnaire survey method, mathematical statistics method and so on with the actual operation. Taking a city as an example, through the literature review, this paper understands the current strategy of national fitness, aiming at the existing problems of national fitness. Under the promotion of innovation 2.0 environment, the fast and effective \"Internet plus\" is used as a means to optimize the supervision and to put forward popularizing strategy and planning path, which has important research value.
Keywords: Internet; National fitness; Fitness; \"Internet plus\"; Innovation 2.0.
Template Security Scheme for Multimodal Biometrics Using Data Fusion Technique
by Gayatri Bokade, Rajendra Kanphade
Abstract: The emerging demand of biometric technology across the globe has given rise to the development of various biometric system involving the multiple traits. The multimodal biometric authentication systems are considered to be more precise and protected as compared to unimodal systems. But when it comes to the stolen or hacked biometric template, even the best designed multibiometric structure fails to prevent intruders entry in the system. In todays digital era where biometric systems are being used for security and access control due to their uniqueness, must ensure to produce highly protected template. This research work proposes Fused Multimodal Principal Component (FMPC) method to generate a secure template by using data fusion technique for three biometric traits face, palmprint and ear. The security of template is tested by implementing various attacks. A feedback routine is instigated to sense and prevent worst case attacks like noise. The proposed method yields promising results.
Keywords: template security; FMPC; attack; face; palmprint; ear.
Design of Electronic Circuit Fault diagnosis based on artificial Intelligence
by Yumei Tao
Abstract: In the large-scale circuit system, the electronic circuit has more components and is prone to fault, so it is necessary to carry out the fault diagnosis design of the electronic circuit to improve the steady-state working ability of the electronic circuit. The thesis proposes a method of electronic circuit fault diagnosis using artificial intelligence algorithm. Intelligent fault diagnosis of electronic circuits first needs to establish a fault signal model of circuit components, and then establish a fault analysis model. The paper uses data mining methods to detect and analyze fault characteristics. The signal detection method is used to analyze and find the parameter state characteristics of the output of the electronic circuit, and the wavelet transform method is used to detect and identify the state. The hardware design of electronic circuit fault diagnosis system based on ARM Cortex-M3 embedded environment is carried out. Experimental research shows that this method has a good judgment function in circuit fault diagnosis.
Keywords: artificial intelligence; electronic circuit; fault diagnosis; fault data mining.
Design of bridge health monitoring system based on b/s mode and soa architecture
by Xiaoyan Jin
Abstract: Aiming at the low integration of bridge health monitoring data management and information planning, the optimal design of bridge health monitoring system is carried out to improve the real-time and accuracy of bridge health monitoring. A design method of bridge health monitoring system is proposed overall framework is analyzed. The system mainly includes information collection module, network module, information adaptive processing module. Bus transmission control module, interface dispatching of bridge health monitoring are bus transmission control of bridge health monitoring system is mode B/S, and analysis of bridge health monitoring information is carried out by using LCD controller. The software development of bridge health monitoring system is realized under SOA architecture, and big data optimizes the management The system test shows that can realize accurately, the integrated processing ability of the bridge operating condition data is better, the information fusion degree is higher, and the system reliability is improved effectively.
Keywords: Aiming at the low integration of bridge health monitoring data management and information planning; the optimal design of bridge health monitoring system is carried out to improve the real-time and accuracy of bridge health monitoring. A design method of bridge health monitoring system is proposed overall framework is analyzed. The system mainly includes information collection module; network module; information adaptive processing module. Bus transmission control module; interface dispatching of bridge health monitoring are bus transmission control of bridge health monitoring system is mode B/S; and analysis of bridge health monitoring information is carried out by using LCD controller. The software development of bridge health monitoring system is realized under SOA architecture; and big data optimizes the management The system test shows that can realize accurately; the integrated processing ability of the bridge operating condition data is better; the information fusion degree is higher; and the system reliability is improved effectively.
Leveraging Bio-Maximum Inverse Rank Method for Iris and Palm Recognition
by Mallikarjuna Reddy A, Reddy Sudheer K, Santhosh Kumar Ch N, K.Srinivasa Reddy
Abstract: Biometrics is vital to recognize and confirm the identity of human beings by estimating and distinguishing the natural qualities that include iris, retinal, face recognition, fingerprint, palm detection, and others. Numerous biometric designs and frameworks are most successful in distinguishing human identities by employing several techniques. In this article, the authors present bi-modular biometric frameworks. For iris and ribbon print, a bi-modular biometric is employed. Wavelet and Gabor-edge channels are employed to separate highlights in various balances. This article aims to present the BMIR (Bio Maximum Inverse Rank) model that is vigorous regarding varieties in scores and other factors of a module. Category support and choice-based strategies are employed to join the magnitudes of these modules. The authors have employed three data sets to carry out the investigation effectively. The investigation shows the accuracy, sufficiency, and appropriateness of the proposed hybrid model when compared with the existing frameworks.
Keywords: Ranking; Iris Recognition; biometrics; knowledge acquisition; neural network.
Local Double Directional Stride Maximum Patterns for Facial expression Retrieval
by Uma Maheswari, Varaprasad G, Viswanadha Raju S
Abstract: Nowadays, face recognition and expression recognition are playing vital role in various applications such as medical field, entertainment, criminal analysis, social media, online business etc. Local texture feature descriptors such as LBP, LTrP, LTP, DBC are usually popular to recognize the faces and expressions as well. In this paper, proposing a new feature descriptor Local Double Directional stride Maximum Pattern to identify the facial expression based on the direction, the pattern generates by calculating the first order derivatives in four directions using DBC, and then second order derivatives are calculated maximum and minimum intensity values in four directions among three pixels in every direction to construct the feature. This helps to discriminates the directional output which is possible to cover in an image. Facial expression recognition and retrieval performance is measured and compared in stand of precision, recall and ARR on the benchmark datasets such as JAFFE, CK+, LFW, FERET etc. with the existing methods.
Keywords: Face recognition; expression recognition; Local Double Directional Stride Maximum Pattern; DBC (Directional Binary Code); LBP (Local Binary Patterns); LTP (Local Ternary Patters); LTrP (Local Tetra Patterns).
DeepVeil: Deep learning for identification of face, gender, expression recognition under veiled conditions
by Ahmad B. A. Hassanat, Abeer Ahmed Al Bustanji, Ahmad S. Tarawneh, Malek Alrashidi, Hani Alharbi, Mohammed Alanazi, Mansoor Alghamdi, Ibrahim S. AlkhazI, V. B. Surya Prasath
Abstract: Biometric recognition based on the full face is an extensive research area. However, using only partially visible faces, such as in the case of veiled persons, is a challenging task. Deep convolutional neural network (CNN) is used in this work to extract the features from veiled-person face images. We found that the sixth and the seventh fully connected layers, FC6 and FC7 respectively, in the structure of the VGG19 network provide robust features with each of these two layers containing 4096 features. The main objective of this work is to test the ability of deep learning based automated computer system to identify not only persons, but also to perform recognition of gender, age, and facial expressions such as eye smile. Our experimental results indicate that we obtain high accuracy for all the tasks. The best recorded accuracy values are up to 99.95% for identifying persons, 99.9% for gender recognition, 99.9% for age recognition and 80.9% for facial expression (eye smile) recognition.
Keywords: Veiled-face recognition; deep learning; convolutional neural; networks; age recognition; gender recognition; facial expression recognition; eye smile recognition.
Recognition of depression patients with electroencephalogram
by Lijie Zhou
Abstract: The recognition of patients with depression is a very important problem, and there are few relevant studies at present. As the application of electroencephalogram (EEG) signals becomes mature in clinical diagnosis, the relationship between EEG and depression has been widely concerned. In this study, firstly, EEG signals were analyzed, then EEG signals were collected for processing and feature extraction, and depression patients were recognized by the support vector machine (SVM) method. The experimental results demonstrated that SVM showed different accuracy in different features, leads, and wavebands. When all leads were used, the accuracy of SVM was the highest. When power spectral density (PSD) was used as the feature, the accuracy of SVM was higher than 70%, and its accuracy on the ? wave was the highest; when activity was used as the feature, the accuracy of SVM was higher than 75%, and its accuracy was the highest on the ? wave. The comparison of random forest (RF) and k-nearest neighbor (KNN) demonstrated that SVM showed the highest accuracy. The results show that the EEG signal based method has a good performance in recognizing depression patients and can be popularized and applied in practice.
Keywords: electroencephalogram signal; depression; support vector machine; feature extraction.
Analysing muzzle pattern images as a biometric for cattle identification
by Worapan Kusakunniran, Anuwat Wiratsudakul, Udom Chuachan, Thanandon Imaromkul, Sarattha Kanchanapreechakorn, Noppanut Suksriupatham, Kittikhun Thongkanchorn
Abstract: Identifying individual animals is important for many reasons of population control, illegal trade prevention, and disease surveillance. This paper focuses on the cattle identification, using biometric-based solution of muzzle images. The proposed method begins with localising muzzle region in each image using the Haar-cascade-based classifier. The scale-invariant feature transform (SIFT) is applied to extract key points of muzzle patterns. Then, SIFT points are split into different clusters/types of muzzle patterns, called bags of muzzle-words (BoM). Finally, the support vector machine (SVM) model is built on BoM as the cattle identifier. The proposed method is evaluated on the published muzzle images dataset of cattle and the collected muzzle image dataset of slaughterhouses and preserved muzzles of swamp buffalo. This article reports the perfect accuracy of 100%. It is also evaluated with the collected dataset of muzzle images of swamp buffalo in the real fields with the reported accuracy of above 90%.
Keywords: cattle identification; muzzle images; animal biometric.
LAHAR-CNN: human activity recognition from one image using convolutional neural network learning approach
by Hend Basly, Wael Ouarda, Fatma Ezahra Sayadi, Bouraoui Ouni, Adel M. Alimi
Abstract: The problem of human action recognition has attracted the interest of several researchers due to its significant use in many applications. With the great success of deep learning methods in most areas, researchers decided to switch from traditional methods-based hand-crafted feature extractors to recent deep learning-based techniques to recognise the action. In the present research work, we propose a learning approach for human activity recognition in the elderly based on convolutional neural network (LAHAR-CNN). The CNN model is used to extract features from the dataset, then, a multilayer perceptron (MLP) classifier is used for action classification. It has been widely admitted that features learned using a CNN model on a large dataset can be successfully transferred to an action recognition task with a small training dataset. The proposed method is evaluated on the well-known MSRDailyActivity 3D dataset. It has shown impressive results that exceed the performances obtained in the state of the art using the same dataset, thus reaching 99.4%. Furthermore, our proposed approach predicts human activity (HA) from one single frame sample which justifies its robustness. Hence, the proposed model is ranked at the top of the list of space-time techniques.
Keywords: human activity recognition; convolutional neural network; CNN; deep learning; daily living activity.
Fingerprint core point detection using connected component approach and orientation map edge tracing approach
by Jincy J. Fernandez, P. Nithyanandam
Abstract: Fingerprint recognition has always played a key role in forensic and commercial applications. The most challenging part of this recognition is to detect the accurate position of the fingerprint's core point. For this, two novel edge-based methods are proposed that improve the accuracy of the fingerprint's core point detection. The first method uses the connected component approach, which helps to clearly distinguish whether the given fingerprint image has a core point or not. Also, it detects the exact location of the core point if it is present. The second method does edge tracing on the orientation map to detect accurate core points in all types of fingerprint images. The proposed works have been compared with some of the existing methods in terms of accuracy, consistency, and execution time. With the samples taken from SDUMLA-HMT database, experimental outcomes have determined its effectiveness by providing an outstanding detection rate.
Keywords: core point; fingerprint recognition; orientation field estimation; edge detection; connected component analysis.
Automatic enrolment for gait-based person re-identification under various view angles
by Imen Chtourou, Emna Fendri, Mohamed Hammami
Abstract: Automatic enrolment constitutes a demanding decision-making practice for person re-identification task but rarely considered in the literature. This paper introduces a new method for automatic enrolment relying on gait analysis. The enrolment problem involves that the gallery database is automatically fed as a new subject is presented. The originality of the proposed method is that in the gallery, a given subject may be represented by several samples. This will improve the re-identification under various view angles. Experiments on CASIA-B database based on accuracy, sensitivity and specificity proved the performance and flexibility of the proposed method.
Keywords: gait; automatic enrolment; person re-identification; view angles.
An empirical evaluation of compression techniques for genome sequences
by M. Muthulakshmi, G. Murugeswari, S.P. Raja
Abstract: Databases of biological sequences are increasing at an exponential rate due to tremendous growth of living organisms. Among all other scientific databases, the size of biological databases is in terabytes. With the advancement in sequencing technologies, each day thousands of nucleotide bases of different organisms are sequenced and submitted to the database worldwide. So, there is a need for compression of biological sequence data to reduce the space required for storage and thereby increase the transmission speed. Three existing sequence compression algorithms namely modified HuffBit, one bit compression and extended American Standard Code for Information Interchange (ASCII) compression algorithms are implemented. The DNA sequence data is obtained from National Center for Biotechnology Information (NCBI) database. The main aim of this paper is to compare and evaluate the performance of existing sequence compression algorithms. Experimental results show that modified HuffBit compress algorithm performs better with an average compression ratio of 3.8.
Keywords: DNA; extended ASCII; genome sequences; modified HuffBit; one bit.
Special Issue on: Biometrics Challenges and Applications
Experimental results on palmvein-based personal recognition by multi-snapshot fusion of textural features
by Mohanad Abukmeil, Gian Luca Marcialis
Abstract: In this paper, we investigate multiple snapshot fusion of textural features for palmvein recognition including identification and verification. Although the literature proposed several approaches for palmvein recognition, the palmvein performance is still affected by identification and verification errors. As well-known, palmveins are usually described by line-based methods which enhance the vein flow. This is claimed to be unique from person to person. However, palmvein images are also characterized by texture that can be pointed out by textural features, which relies on recent and efficient algorithms such as Local Binary Patterns, Local Phase Quantization, Local Tera Pattern, Local directional Pattern, and Binarized Statistical Image Features (LBP, LPQ, LTP, LDP and BSIF, respectively), among others. Finally, they can be easily managed at feature-level fusion, when more than one sample can be acquired for recognition. Therefore, multi-snapshot fusion can be adopted for exploiting these features complementarity. Our goal is to show that thisrnis confirmed for palmvein recognition, thus allowing to achieve very high recognition rates on a well-known benchmark data set.
Keywords: Palmvein Recognitioon; multi-snapshot fusion; local dense descriptor.
Euclidean Distance versus Manhattan Distance for Skin Detection using the SFA Database
by Ouarda Soltani, Souad BENABDELKADER
Abstract: Skin detection is very challenging because of the differences in illumination or cameras characteristics, or the range of skin colors due to different ethnicities, and many other variations. Among important contributing tools to developing new methodologies more effective and accurate for skin color detection we could easily identify color human skin databases specifically designed to assist research in the area of face recognition. One of these is the recently built SFA database that showed high accuracy for segmentation of face images. In this particular context, the approach described in this paper exploits skin and non-skin samples provided by SFA for skin segmentation on the basis of the well-known Euclidean and Manhattan distance metrics. Most importantly, the scheme proposed tries to segment facial color images inside or outside SFA by means of skin samples belonging to SFA. Simulation results in both SFA and UTD color face databases indicate that detection rates higher than 95% can be achieved with either measure.
Keywords: Skin segmentation; Skin color detection; Euclidean distance; Manhattan distance.
Identifying Age Group and Gender Based on Activities on Touchscreen
by SOUMEN ROY, Utpal Roy, Devadatta Sinha
Abstract: Predicting users age vis-
Keywords: Personal Traits; Keystroke Dynamics; Touch Dynamics; XGBoost; LOOCV; LOUOCV; Smartphone Sensors.
A Robust and Efficient Fingerprint Minutiae Extraction in Post-Processing Algorithm
by R. Anandha Jothi, V. Palanisamy
Abstract: Minutiae extraction based fingerprint recognition is important for person authentication. However its own drawbacks and challenges such as missing and spurious minutiae which leads to a slight degree of inaccuracy. Herein an attempt has been made to overcome these challenges and achieve a higher degree of accuracy for minutiae extraction. In pre-processing the fingerprint image is subjected to frequency domain filter (FDF) and uses Fast Fourier Transform (FFT) to improve the fingerprint image by conjoining some faultily broken points on ridges, removing some false connections between ridges and improves the appearance of the ridges by filling up small holes. Further the filtered fingerprint ridge and its valley are subjected to binarization and thinning process. Following this Rutovitz Crossing Number (CN) method is used to ensure the preservation of true minutiae and removal of false minutiae. In post-processing Grahams scan algorithm based convex hull filtering technique (CHFT) is effectively used to eliminate the bogus minutiae lying on the boundary of the fingerprint. It is observed that, the CHFT can be easily superposed on fingerprint image and the precision of the filtered minutiae verified visually. After post-processing, experimental results of ridge and valley based minutiae were analyzed individually and compared. It clearly revealed that the accuracy rate and goodness of index (GI) were better in ridges whereas the computational speed of post-processing was much faster in valley.
Keywords: Minutiae extraction; Fast Fourier transform; convex hull; Graham’s scan; Ridge; Valley.
FACE DETECTION AND RECOGNITION SYSTEM BASED ON HYBRID STATISTICAL, MACHINE LEARNING AND NATURE BASED COMPUTING
by Vinodini Ramamurthy, Karnan M.
Abstract: Face detection becomes an important task carried out in biometric based security system and identification application. This paper presents the detailed investigations on different methods suffer from accuracy and computational complexity used for the face detection and recognition. The face detection and recognition with high performance ratio for face detection and recognition is achieved in the methods investigated. The reduction of complexity can happen at any stages of the face recognition like preprocessing, segmentation, feature extraction, recognition etc. The proposed method presented in this paper is based on PCA (principle component analysis), SVM (support vector machine), K-nearest neighbor (KNN) and ACO (ant colony optimization). The detail investigation of the proposed method is made and is compared with the existing methods. From the performance it can be observed that the proposed method is better in performance when compared to other methods.
Keywords: Face detection; recognition; PCA; SVM; ACO; segmentation; feature extraction; classification.
Special Issue on: Biometrics, Deep Learning and Sentiment Analysis
Cascading failure of complex networks under degree-based attack
by Yan Liu, Jie Yang, Peng Geng
Abstract: This paper summarises the research status of cascading failure in communication networks, power grids and independent networks. Considering the load and capacity of nodes, a cascading failure model of complex networks for degree-based attack is established. In this model, a new measurement method is proposed. This method is a normalised index to measure the impact of cascading failure, which is used to study the global consequences of different types of node failure. By attacking the nodes with higher and lower degree values, the effects of higher-degree-based attack strategy and lower-degree-based attack strategy on network robustness are discussed. The simulations show that when certain conditions are met, attacking nodes with a lower degree (lower-degree-based attack) can cause greater damage. This result breaks the conventional thinking and provides a reference for the protection of important nodes in the real complex networks.
Keywords: cascading failure; complex networks; degree-based attack; higher-degree-based attack; lower-degree-based attack.
Special Issue on: Machine Learning Algorithms for Biometrics
The model of fast face recognition against age interference in deep learning
by Yuzhe Zhang, Peilin Wu, Jinhui Zhao, Hao Feng, Rongtao Liao
Abstract: In order to overcome the low recognition efficiency of traditional anti-age face recognition methods, the paper proposes a new anti-age-disturbing face recognition modeling method based on deep learning. Build a standard face recognition information database and use this as the matching standard for face recognition. Construct a deep learning convolutional neural network, install the propagation process and training strategy of deep learning, build an age discrimination model, build a loss function and an objective function and solve it. On this basis, the facial features are extracted, after matching with the data in the standard database and similarity calculation, the final rapid facial recognition result is obtained. Experimental results show that the highest recognition accuracy of the designed face recognition model is 99.2%, and the recognition speed of the designed model is faster.
Keywords: deep learning; convolution neural network; anti age interference; face recognition model;.
Arm Movement Recognition Of Badminton Players In The Third Hit Based On Visual Search
by Ling Liu
Abstract: In order to overcome the problems of low recognition accuracy, high repetition rate of corner points and long recognition time of badminton players' arm movements in the third beat, a new recognition method based on visual search is proposed in this paper. This method uses visual search technology, uses binocular visual camera to record video, uses image stabilization technology to implement histogram equalization processing on the recorded visual image, obtains the visual image after noise elimination, uses scale invariant feature conversion algorithm to collect image sift corner features, combines with support vector machine algorithm to construct feature recognition model, and realizes the third shot arm movement of badminton players distinguish. The experimental results show that the signal-to-noise ratio (SNR) of the collected visual images is less than 10dB, the repetition rate of corners is always lower than 60%, and the recognition time is basically controlled within 0.8s.
Keywords: Visual search; badminton players; arm movement; recognition; visual image; corner features.
Multi pose facial expression recognition based on convolutional neural network
by Yongliang Feng
Abstract: In order to overcome the problems of low expression similarity and low recognition rate in multi pose facial expression recognition, a new multi pose facial expression recognition method based on convolutional neural network is proposed. The convolution layer is constructed directly by Gabor wavelet with fixed weights, and the full connection layer is constructed by support vector machine (SVM). The structure of convolution neural network is determined by matching growth rules, and the network parameters are trained by back-propagation algorithm. Adaboos algorithm is used to cut facial expression, gradient integral projection and dual threshold binarization are used to locate eyes. The scale normalization and gray scale normalization are used to realize multi pose facial expression recognition. The experimental results show that the highest expression similarity is 98.43%, the recognition rate is close to 100% under different rotation angles, and the recognition rate is as high as 99.96% for different expressions.
Keywords: Convolution neural network; Multi pose; Facial expression; Eecognition; Adaboos algorithmrn.
Facial feature localization and subtle expression recognition based on deep convolution neural network
by Qiaojun Li, Peipei Wang
Abstract: In order to solve the problems existing in traditional face recognition methods, such as low accuracy of face feature location, poor accuracy of subtle expression recognition and long recognition time, a face feature localization and subtle expression recognition based on deep convolution neural network is proposed. The principle of deep convolution neural network is analyzed, and the feature extraction of human face is placed in convolution layer and pooling layer. The foreground and background entropy of face image are obtained by binarization method of face image, and optical flow characteristics of all positions of frame image are obtained by using deep convolution neural network, and the recognition of facial subtle expression is completed. The experimental results show that the accuracy of the proposed method is up to 98%, the recognition accuracy of facial expression is high, and the recognition time is short .
Keywords: deep convolution; neural network; facial feature localization; subtle expression; recognition.
Research on Adaptive Conversion of AI Language based on Rough Set
by Yuping Fang, Da Fang
Abstract: In order to solve the problems of high complexity and low computational efficiency in traditional artificial intelligence language conversion methods, an adaptive artificial intelligence language conversion method based on rough set is proposed. Artificial intelligence language (AI) preprocessing is realized by pre-emphasizing, adding window, frame processing and endpoint detection. The attribute reduction algorithm based on rough set theory is used to select the features of ARTIFICIAL intelligence language. The dimension of input feature vector is reduced. The experimental results show that after feature extraction, the computational efficiency is obviously improved, and the efficiency of the proposed method is the highest, averaging close to 100%. Compared with the traditional method, the complexity of the proposed method is lower, and the average complexity is 1.68% during the 10 experimental iterations.This method simplifies the adaptive conversion process of ARTIFICIAL intelligence language and has high computational efficiency.
Keywords: Rough det; AI language; Adaptive conversion; Feature selection; Redundant information deleting.
Face feature tracking algorithm for long-distance runners based on multi region fusion
by Li Wan
Abstract: In order to overcome the problems of high error rate and poor tracking effect of traditional algorithms, a multi-region fusion based feature tracking algorithm for long-distance runners was proposed in this paper. Firstly, the multi-region template voting strategy is adopted to classify and obtain face features by dividing face feature similarity threshold in different regions through regional feature similarity classification. Then the mean-Shift tracking algorithm was used to complete the target object modeling, and the pap coefficient was used as the evaluation standard of model similarity measurement, and the face features were tracked through iterative operation. Experimental results show that the recognition accuracy of this algorithm is higher than 92% in different situations, and the tracking error of the center position is always below 20 pixels in different angles and complex environments, which fully proves the effectiveness of this algorithm.
Keywords: Multi region fusion; Long-distance runner; Feature classification; Face feature tracking; Mean Shift tracking algorithm.
Multi-information fingerprint identification method based on Interactive Genetic Algorithm
by Yuansheng Liu
Abstract: In order to overcome the shortcomings of traditional fingerprint identification methods, such as low computational efficiency and high false classification rate, a new multi-information fingerprint identification method based on interactive genetic algorithm is proposed. Firstly, the multi-information fingerprint image is preprocessed to extract the feature points. Then, combined with the interactive genetic algorithm, the rotation angle of the two fingerprint images is determined. At the same time, the fingerprint offset code is coded, and the matching setting function is set. The matching degree of the two fingerprints is determined by the interactive genetic algorithm, which effectively realizes the multi-information fingerprint identification. Finally, the simulation experiment is carried out. The experimental results show that the proposed method can effectively reduce the false classification rate, reduce the average matching time of fingerprint image, and improve the operation efficiency. The minimum error rate is only 1.02%.
Keywords: Interactive genetic algorithm; Feature point extraction; Multi-information fingerprint identification; Offset coding.
Dynamic facial expression recognition of sprinters based on multi-scale detail enhancement
by Xiang CAO, Pengfei Li
Abstract: In order to solve the problems of low average gradient and long recognition time in traditional facial expression recognition method, a multi-scale detail enhancement method for facial dynamic expression recognition of sprint athletes is proposed. A principal component analysis method was used to establish the facial expression feature subspace of sprinters, to project and reduce the dimension of the facial dynamic expression feature vector of sprinters, and to obtain the low frequency information and high frequency information of the facial image of sprinters by bilateral filtering. The multi-scale details of expression are enhanced by using side suppression network model and improving image S curve. The feature vector of facial dynamic expression is input into support vector machine to recognize the facial dynamic expression of sprinter. Experimental results show that the average value of annoying gradient is about 98 and the shortest time s. is about 1.9 .
Keywords: Multi-scale; Image detail enhancement; Gabor wavelet transform; Feature vector; Expression recognition; Support vector machine.
Research on automatic recognition method of basketball shooting action based on background subtraction method
by Linzhu Li, Kun Wang
Abstract: In order to overcome the problems existing in the existing methods, such as high rate of false recognition, high rate of missing recognition and low rate of recognition, an automatic recognition method of basketball shooting action based on background subtraction method is proposed. The background information of the target image is obtained by the background subtraction method to improve the clarity of the whole contour of the moving object. A certain number of video frames are extracted, and the background in the image is effectively separated. The two-dimensional Fourier transform is used to balance the video frames to obtain the foreground target information and the background of the video sequence, so as to complete the automatic recognition of basketball shooting action. Experimental results show that the proposed method can effectively reduce the rate of false recognition and missed recognition, and improve the recognition rate.
Keywords: Background subtraction method; basketball shooting; action automatic recognition; optical flow method; error rate.
Recognition Method of Unspecified Face Expressions Based on Machine Learning
by ZheShu Jia, DeYun Chen
Abstract: Traditional face recognition methods usually complete facial expression recognition for designated faces, and the pixel set at the edge of face image is chaotic, which leads to poor accuracy of unspecified facial expression recognition. In order to improve the accuracy of unknown facial expression recognition, a method of unknown facial expression recognition based on machine learning is proposed. The feature detection model of unspecified facial expressions is constructed, and the features are divided into regional blocks. Fusion block feature information. Establishing spatial feature projection model, weighting feature information entropy, extracting statistical features and edge information entropy features, reorganizing features and matching edge pixel sets, and completing the recognition of various facial expression features. Experimental results show that the accuracy of this method is significant, reaching 1, which effectively improves the recognition efficiency and anti-interference performance.
Keywords: Machine learning; unspecified person; facial expression; recognition; feature extraction; information enhancement.
Research on leg posture recognition of sprinters based on SVM classifier
by Yang HE
Abstract: In order to overcome the problems of low recognition rate, high time consumption and high misclassification rate caused by the difficulty in obtaining the global motion pattern information of sprinters in traditional posture recognition methods, a leg pose recognition method based on SVM classifier is proposed. Using multivariate statistical model to denoise the sprint image, the effective leg movement pattern information of sprinters is extracted. In the SVM classifier, the samples are divided by decision function to realize the recognition of sprinters' leg posture. In order to verify the effectiveness of the method in this paper, a comparative experiment is designed. Experimental results show that the recognition rate of the proposed method is more than 90%, the time consumption of recognition process is always less than 0.5s, and the misclassification rate of leg pose features is always below 5%, which fully demonstrates the high recognition performance of the method.
Keywords: SVM classifier; Sprint; Wavelet transform; Feature vector; Leg pose recognition; Feature extraction.
Automatic recognition of javelin athletes throwing angle based on recognizable statistical characteristics
by Zhe Dong, Xiongying Wang
Abstract: In order to overcome the problem that the traditional recognition method has poor statistical performance on the regularity of body feature data before recognizing the throwing angle, which leads to the deviation of javelin flight trajectory judgment results, this paper proposes an automatic recognition method of javelin athletes' throwing angle based on the recognizable statistical characteristics. Firstly, the technical characteristics of javelin throwers of different genders are extracted by using the statistical process of distinguishing features. Then, the angle of recognition equipment is calibrated and the position of trigger signal is combined to realize the automatic recognition of javelin throw angle. Experimental results show that: the javelin flight trajectory identified by this method is the closest to the actual trajectory, and the recognition accuracy of the throwing angle can reach more than 98%. It shows that the method can effectively realize the accurate recognition of javelin athletes throwing angle.
Keywords: Recognizable statistical characteristics; javelin movement; throwing angle recognition; two-dimensional image; trigger signal position; recognition accuracy.
Face feature tracking algorithm of aerobics athletes based on Kalman filter and Meanshift
by Shu Yang
Abstract: In order to solve the problems of low accuracy and long time-consuming in face image tracking of Aerobics Athletes in traditional methods, a face feature tracking algorithm based on Kalman filter and meanshift is proposed. Three frame difference method is used to extract the colour features of Aerobics Athletes' face images, measure the geometric feature similarity of Aerobics Athletes' face images, calculate the gray value of local images of Aerobics Athletes' face features, and match corner features by NCC matching algorithm. The Kalman filter method is introduced to denoise the different pixels of the feature image, and the mean shift of the Aerobics Athletes' face features is obtained by means of the mean shift algorithm to realise the tracking of the Aerobics Athletes' face features. The experimental results show that the tracking accuracy of the proposed method is up to 97%, and the shortest tracking time is about 1.5 s.
Keywords: Multi region fusion; aerobics athletes; image resolution; background modelling; feature tracking;.
Special Issue on: Biometrics in Smart Cities Techniques and Applications
Multidimensional distribution data association algorithm based on DNAzyme
by Yingying Liu, Xixi Li, Shaofeng Rong, Shimin Guan, Baoguo Cai, Shuo Zhang
Abstract: The distribution function of the current multi-dimensional distribution data association algorithm does not have timeliness, which causes the data to generate interference signals during the distribution process, and there are correlation trajectory errors. Based on DNAzyme, a multi-dimensional distribution data association algorithm is proposed. Use the classification characteristics of mostly distributed data to extract the characteristic parameters of the data, and at the same time strengthen the internal storage structure of different data to ensure the safe storage of data, connect the signal between the sensor and the data in time, establish a good data communication relationship, and select target tracking The way to estimate the distribution data state, obtain the information that matches the correlation law, and execute the correlation algorithm operation, adjust the distribution performance of the distribution data, and narrow the gap between the measurement data correlation trajectory and the real data correlation trajectory. Experimental results show that under the same parameter conditions, the proposed algorithm has stronger timeliness and the associated trajectory error rate is smaller.
Keywords: DNAzyme; multidimensional distribution data; data association; association algorithm.