Abstract: In order to realize the rapid identification of different cuts of chilled lamb at different storage times by using mobile phones, images of back, hind leg and front leg were collected over a storage time of 12 days. Color features such as the mean value in different color spaces and the color moments in the red, green and blue (RGB) color space were extracted and analyzed for significant differences, and 7 color features with significant differences among meat cuts were obtained. Considering the differences in color features among meat cuts as well as different color spaces, four combinations of color features were selected as the inputs for the extreme learning machine (ELM), back propagation (BP) and support vector machine (SVM) models, respectively, and modeling analysis and comparison were carried out to classify different lamb cut. The results showed that upon using each color feature combination as the model input, the classification performance of the BP model established was better than that of the SVM and ELM models. When the 12 color mean features were chosen as the input, the BP model showed the best classification performance with a discrimination accuracy of 96.13%, 95.11% and 91.44% for the training, cross-validation, and test set, respectively, allowing rapid discrimination of lamb cuts at different storage times. The research provides a theoretical basis and technical support for future development of mobile phone applications to rapidly discriminate chilled lamb cuts at different storage times.

Key words: chilled lamb; different meat cuts; mobile phone image; storage time