Detection and identification of foreign bodies in conditioned steak based on ultrasound imaging

• Published in: FOOD SCIENCE AND TECHNOLOGY RESEARCH Vol. 30; no. 3; pp. 269 - 280
• Main Authors: Li, Chen, Niu, Zeng, Zuo, Min, Wang, Tianzhen, Zou, Xiaobo, Sun, Zongbao
• Format: Journal Article
• Language: English
• Published: Tsukuba Japanese Society for Food Science and Technology 01.01.2024; The Japanese Society for Food Science and Technology; Japan Science and Technology Agency
• Subjects: Artificial neural networks; Bones; conditioned steak; Crush tests; Discriminant analysis; ELM; Feasibility studies; Foreign bodies; foreign body; Hazard identification; Imaging; Machine learning; Manufacturing industry; Pattern recognition; Texture recognition; Ultrasonic imaging; Ultrasonic testing; Ultrasound
• ISSN: 1344-6606; 1881-3984
• DOI: 10.3136/fstr.FSTR-D-23-00068

Conditioned steak is easily contaminated by foreign bodies, such as iron sheets, glass, and crush bones in the manufacturing processes, posing hidden safety hazards to consumers. In this study, the feasibility of using ultrasonic imaging to detect and identify foreign bodies in conditioned steaks was investigated. Firstly, the ultrasonic imaging data of foreign bodies was collected. Four discriminant models among them linear discriminant analysis (LDA), and extreme learning machine (ELM) were established, and based on the texture values of the smallest circumscribed rectangular area of the foreign bodies, the type was identified. The foreign bodies were then extracted by gray–level co–occurrence matrix (GLCM). The detection rate of foreign bodies was 97.78 %, meanwhile ELM showed the highest accuracy of recognition rate of 76.67 %. The results showed that ultrasound imaging technology could be used to detect foreign bodies in the conditioned steak and to identify the type of foreign body via pattern recognition.