International Journal of Signal and Imaging Systems Engineering (7 papers in press)

Regular Issues

• On The Performance of a Fuzzy Variable Structure Satellite Attitude Controller Under Sensor and Actuator Uncertainties  
  by Bilgehan Erkal 
  Abstract: Controlling the attitude of a satellite with high accuracy and stability under both sensor and actuator delay is a great problem. It is possible to correct for errors, but a robust controller is more preferable. In this study, the attitude of a 3-DoF satellite model incorporating uncertainties (delays in handling both sensors and actuators) is controlled using a suitably designed Integral Fuzzy Variable Structure Controller (IFVSC). The attitude control accuracy of the IFVSC is evaluated and compared to other reference controllers (one is a PID Controller and the other is a Loop Shaping Controller). IFVSC is found to perform well with Td=0.2s sensor and actuator data delay.
  Keywords: Fuzzy Variable Structure Control; Loop Shaping Controller; Satellite Attitude Control; Sensor Data Delay; Actuator Misplacement.
  DOI: 10.1504/IJSISE.2020.10032027
   
  
• VAD, feature extraction and modeling techniques for speaker recognition - a review  
  by B.G. Nagaraja 
  Abstract: This article reviews an automatic speaker recognition technology, with an emphasis on state-of-the-art voice activity detection (VAD), feature extraction and speaker modeling techniques that have emerged during the last few years. Researchers in the field of speaker recognition have made a few attempts to recognize the speaker in the language mismatch environment and limited data condition. To address robustness issues, we also elaborate language mismatch and limited data speaker recognition. Further, this paperrnidentified some issues with the existing speaker recognition systems and also investigated areas of possible improvements in speaker recognition field. We conclude the paper with a discussion on the possible future directions.
  Keywords: VAD; speaker identification; speaker verification; language mismatch; limitedrndata; multilingual; features; modeling techniques.
  
  
• Robust speaker recognition based on biologically inspired features  
  by Youssef Zouhir, Ines BEN FREDJ, Kaïs Ouni 
  Abstract: This paper propose two speech parameterization techniques for noise-robust speaker recognition: the Normalized Gammachirp Cepstral Coefficients (NGCC) and the Perceptual Linear Predictive normalized Gammachirp (PLPnGc). These techniques are based on a biologically inspired auditory model which simulates the cochlea spectral behaviour. The Gaussian Mixture Model-Universal Background Model (GMM-UBM) based speaker modelling is considered in automatic speaker recognition system. The performances are evaluated in clean and noisy environments using Timit, Aurora and Demand databases. The experimental results in noisy environments showed that the biologically inspired feature extraction techniques give a better recognition rate than state-of-the-art techniques.
  Keywords: Auditory filter model; Biologically inspired features; Normalized GammachirprnCepstral Coefficients (NGCC); Perceptual Linear Predictive normalized Gammachirp (PLPnGc); Gaussian Mixture Model-Universal Background Model (GMM-UBM); Robust speaker recognition;.
  
  
• Lossless and near lossless compression of images with sparse histograms  
  by Souha Jallouli, Sonia Zouari, Nouri Masmoudi, Atef Masmoudi 
  Abstract: Histogram sparseness is a characteristic which is expected by most of the lossless compression algorithms. In fact, they have been designed mainly to process continuous-tone images. Images with sparse histograms are characterized by a small number of pixels\'intensity levels compared to the one implied by the nominal bit depth. Moreover, the active levels may spread throughout the nominal intensity range since they do not occupy all the continuous range. However, the compression efficiency of most of lossless image encoders is severely affected when handling sparse histogram images. This paper presents an analyze of the impact of histogram sparseness on lossless image compression standards in case of predictive coding. In addition, to improve the compression performance for sparse histogram images, a new preprocessing technique is proposed. The method takes advantage of the high likelihood between neighboring image blocks. The initial image is divided into blocks and for each image block, the proposed method associates the most reduced set representing its active symbols and makes the histogram dense. This technique is efficient without implying any modification of the basic code of the state-of the art lossless image compression techniques. We show experimentally that the proposed method outperforms JPEG-LS, CALIC and JPEG 2000 and achieves lower bitrates.
  Keywords: Lossless image compression; sparse histogram; histogram packing; predictive coding; JPEG-LS; CALIC; JPEG 2000.
  
  
• HMM-GMM based Amazigh Speech Recognition System  
  by E.L. Ouahabi Safâa, MOHAMED ATOUNTI, Bellouki Mohamed 
  Abstract: this study presents conception and realisation of an automatic independent speech recognition system using Hidden Markov Model (HMM). The system recognizes 33 letters in Amazigh language. System is found well performed and can identify the Amazigh spoken letters at 88, 44% recognition rate, which is well acceptable rate of accuracy for speech recognition. The tests were taken based on the Hidden Markov Model and Gaussian mixture distributions. Hidden Markov Toolkit (HTK) has been used in implementation and test phases. The Word Error Rate (WER) came initially to 29,41 and reduced to about 11,52% thanks to extensive testing and change of the recognitions parameters.
  Keywords: Automatic Independent Isolated-Letters Speech Recognition System; Amazigh Language; Hidden Markov Model (HMM); Gaussian Mixture Models (GMMs); Hidden Markov Model Toolkit (HTK); Word Error Rate (WER).
  
  
• Image Colour Edge Detection using Hypercomplex Convolution  
  by Rawan Zaghloul, Hazem Hiary 
  Abstract: Quaternions are considered for colour image edge detection. Most work on quaternions is based on a linear quaternion system (LQS) which applies multi-directional kernels (horizontal, vertical, and diagonal) using hypercomplex convolution, each kernel producing an edge map for a specific direction, and the final result is a combination of these maps. This paper introduces a new colour image edge detection filter based on LQS convolution. The process starts by applying quaternion convolution with the proposed filter, and then generating the final edge map by computing the magnitude of the result. The proposed filter is able to highlight both colour and greyscale edges in multiple directions using a single LQS convolution pass. The validity of the proposed filter is demonstrated, and its performance is supported experimentally through a set of comparisons with state-of-the-art methods.
  Keywords: quaternions; colour image; edge detection; hypercomplex convolution; multi-directional kernels; Linear Quaternion System (LQS).
  
  
• Performance Analysis of Fusion Based Brain Tumour Detection Using Chan-Vese and Level Set Segmentation Algorithms  
  by Rajeh Babu Katta, Naganjaneyulu P. V., Sathya Prasad K 
  Abstract: Brain tumour shortens the life expectancy of the diseased if not identified at early stages. Accompanied by variety of segmentation algorithms, MRI has been widely used as one of the identification procedures. But no single technique is commonly accepted for accurate segmentation that correlate with pathological studies. This paper highlights the effectiveness of CNN fusion followed by Chan-Vese active contour based segmentation intended for the detection of brain tumour and compares it performance with other contemporary approaches using various metrics.
  Keywords: Fuzzy C-Means; K-Means; CNN; CT; NSCT; MWGF; GFF; Chan-Vese; Level Set.