International Journal of Product Development (7 papers in press)

Regular Issues

• A Biomimetic Feature Extraction Method for Product Packaging Colour Based on Flower Pollination Algorithm  
  by Ting Zhang, Rui-zhi Liu, Wei Li 
  Abstract: To improve the accuracy and recall of product packaging colour bionic feature extraction, this paper proposes a product packaging colour bionic feature extraction method based on flower pollination algorithm. Firstly, smooth the product packaging image, extract superpixel blocks from the image, and complete intelligent image segmentation. Then, using the flower pollination algorithm to simulate self pollination, the conversion probability is introduced to conduct a global search for colour biomimetic features and determine the colour biomimetic features; Finally, using the sigmoid function, the determined colour biomimetic features are discretized and extracted. If the feature value is 0,the feature is not extracted; If the value is 1, the feature is extracted to complete the colour biomimetic feature extraction. The results show that the extraction accuracy of the method in this article can reach 99.25%, the extraction recall rate can reach 96.2%, and the product packaging colour biomimetic feature effect is good.
  Keywords: flower pollination algorithm; Color biomimetic feature extraction; Sigmoid function; Product packaging; Swarm intelligence optimisation algorithm.
  DOI: 10.1504/IJPD.2025.10068635
   
  
• Error Detection of Industrial Design Product Appearance Dimensional Based on Machine Vision  
  by Hua Song 
  Abstract: Aiming at the problems existing in current methods, such as high false detection rate, low signal-to-noise ratio of image edges and high cost of sub-pixel matching, an error detection method of industrial design product appearance dimensional based on machine vision is proposed. The fuzzy algorithm is used to extract the edge of industrial design product appearance image, and the sub-pixel point matching is carried out after determining the amplitude change of sub-pixel points in the edge image. According to the pixel coordinates and image parallax of the appearance image, the standard threshold of the appearance image dimensional of industrial design products is set, and the appearance dimensional image to be detected is compared with the standard threshold of the image dimensional to realize error detection. Test results show that the proposed method has low false detection rate, high signal-to-noise ratio of image edge and low cost of sub-pixel point matching.
  Keywords: Machine vision; Industrial design products; Appearance dimensional; Error detection; Amplitude; Sub-pixel point matching.
  DOI: 10.1504/IJPD.2025.10068647
   
  

Special Issue on: The Future of New Product Development in Digital Markets PART 1

• A Precision Marketing Method for Digital Product Big Data Based on User Generated Content  
  by Jing Liu, Yiwen Ruan, Jia Lin 
  Abstract: In order to improve the marketing accuracy and user satisfaction of digital product big data, a precision marketing method based on user generated content for digital product big data is proposed. Firstly, vectorise the user generated evaluation text, digital product category text and image information of digital product descriptions. Secondly, convolutional fusion is performed on the text comprehensive features and image features of digital products. Finally, construct a digital product user interest model based on the level of user interest. Using tag weights to construct a precise marketing function for digital product big data. The experimental results show that compared with existing marketing methods, this paper method can improve the marketing accuracy of digital product big data, while also enhancing user satisfaction.
  Keywords: User generated content; Digital products; Big data precision marketing; User interest model.
  DOI: 10.1504/IJPD.2024.10067928
   
  
• Personalised Push Method for Sports Goods Purchase Information in the Context of Marketing  
  by Ziya Wang, Fei Gao 
  Abstract: To solve the issues of low satisfaction and low UV click rate in the existing personalised push methods of commodity purchase information, this paper proposes a personalized push method for sports goods purchase information in the context of marketing. By utilising crawler technology, an online shopping platform users' behaviour dataset for purchasing sporting goods is established. A generalised hierarchical tree of user portrait attribute tags is constructed, and the browsing time of sporting goods containing a certain tag is calculated to allocate the membership degree of sporting goods and mine user interest preferences. The Pearson similarity algorithm is applied to construct the similarity matrix for personalised push, enabling personalised push of sporting goods purchase information. Experiments demonstrate that the user satisfaction rate using this method consistently remains above 88%. Furthermore, the highest UV click rate achieved is 7.56%, indicating a successful push effect.
  Keywords: Sports goods; User portrait; Generalised hierarchical tree; Membership degree; Pearson similarity.
  DOI: 10.1504/IJPD.2024.10067945
   
  
• Research on Image Detail Enhancement of Cultural and Creative Product Packaging Design based on Improved Guided Filtering  
  by Jin Yan 
  Abstract: In order to avoid image distortion caused by excessive processing during image detail enhancement, an image detail enhancement method for cultural and creative product packaging design based on improved guided filtering was proposed. Obtain the packaging design image of cultural and creative products, use the improved non local mean filtering algorithm to denoise the image, and repair the bad points in the image to protect important details; The multi-scale guidance filter is introduced to improve the guidance filter, and the multi-scale guidance filter is used to enhance the details of the modified image. The experimental results show that the minimum peak signal-to-noise ratio of the image enhanced by this method can reach 53.2db, the entropy value is 0.94, and the minimum average contrast of the image can reach 0.85, indicating that the image processed by this method has high quality and high detail retention rate.
  Keywords: Cultural and creative product packaging; Design image; Detail enhancement; Improved guided filtering; Lifting wavelet.
  DOI: 10.1504/IJPD.2024.10067947
   
  
• Fuzzy Comprehensive Evaluation of Product Marketing Management Performance under the Background of Data Driven  
  by Ming Yang 
  Abstract: In order to overcome the problems of low weight and score values, large fluctuations in indicator membership, and low credibility in traditional methods, this paper designs a new fuzzy comprehensive evaluation method of product marketing management performance under the background of data driven. Determine financial performance indicators, innovation performance indicators, and market competition performance indicators. Utilizing data driven technology to mine and clean performance indicators for product marketing management, combined with dimensionless evaluation indicators and consistency calculation to complete indicator data preprocessing. Build a fuzzy comprehensive evaluation model for product marketing management performance, input preprocessed indicator data into the model, and obtain evaluation results. The test results show that the weight and score values of this method are high, the fluctuation of the membership degree of the evaluation index is low, and the credibility of the comment results is high.
  Keywords: Data driven; Product marketing; Management performance; Fuzzy comprehensive evaluation; Dimensionless; Data driven technology.
  DOI: 10.1504/IJPD.2024.10067948
   
  
• Speech Recognition Interaction Control Method for Smart Home Based on Natural Language Processing  
  by Shuqing Wu 
  Abstract: In order to solve the problems of high speech recognition error rate, low success rate, and high latency in traditional methods, a new speech recognition interaction control method for smart home based on natural language processing is proposed. A first-order FIR high-pass digital filter is used for pre-emphasis of smart home voice, signal framing is achieved by adding a Hamming window, and smart home voice signal enhancement is achieved by combining a linear filter. An improved DTW algorithm is used to recognise the smart home voice. Based on the smart home voice recognition result and natural language processing technology, interactive control instructions are determined and input into a fuzzy self-tuning PID controller to achieve smart home voice recognition and control. Experimental results show that the mean recognition error rate of this method is 5.36%, the mean success rate is 97.52%, and the delay is below 0.6s.
  Keywords: Natural language processing; Smart home; Speech recognition; Interactive control; Linear filter; Improved DTW algorithm.
  DOI: 10.1504/IJPD.2024.10067949