Abstract: In this paper, a method for extracting browning features from cooked shrimps based on a simplified color model was proposed. A large number of images of normal cooked shrimps and melanotic cooked shrimps in the dynamic production line were obtained by online image acquisition equipment, and the region of interest of cooked shrimps was extracted by using the grey-scale difference method and morphological operations, and the browning features of cooked shrimps were extracted by three color spaces; a color model was constructed for 151 images of normal shrimps and 66 images of melanotic shrimps in the training set, and the percentage of browning areas of cooked shrimps were used as the benchmark condition to determine whether browning occurred in cooked shrimps, and were graded. The results showed that the RGB color model with the B-component range of 0~77 was the most effective, and the recognition rate could reach 91.36 ms/each of the 116 normal and 50 browned shrimp images, which reduced the complexity of the algorithm from O(n²) to O(n) and improved the recognition accuracy by 7.3% to 92.77% compared with the traditional binary linear discriminant analysis method. The method has greater practical advantages and application prospects, providing performance support for the application of machine vision technology to the online non-destructive inspection of the appearance quality of South American white shrimp.