RESEARCH AND EXPERIMENTAL STUDY ON THE CLASSIFICATION OF SEAFOOD SNAILS BASED ON MACHINE VISION

Aiming at the current problems of high labour intensity and costs associated with the manual sorting of sea freshwater snails, we proposes a male and female classification model using DPO-SVM. The texture features of the shell intervals were extracted by grey scale covariance matrix analysis, and SVM was used as the classifier to compare the effectiveness of different combinations of texture features. It was concluded that the classification effect of using the energy, entropy, and contrast was the best. To optimize the SVM parameters c and g, the DPO algorithm, based on PSO and WOA algorithms, was introduced. The performance of DPO-SVM was tested and compared with the standard SVM, PSO-SVM, and WOA-SVM models. The results demonstrate that DPO-SVM significantly improves, with overall accuracy rising from 85% to 100%, representing a 15% over the basic SVM model. Additionally, DPO algorithm improves the optimisation seeking performance of the single-species population optimisation algorithm, increasing the best fitness from 95.26 to 98.68 (a 3.47% improvement) and reducing the number of iterations needed to achieve optimal fitness from 14 to 6, a 57.14% decrease ration. The research provides a valuable technical reference for the male and female classification of seafood conchs in automatic sorting device.