Research on kiwifruit grading detection system based on depth limit learning machine

Abstract

Aiming at the low recognition rate and low accuracy of traditional kiwifruit grading detection, a feature extraction method based on machine vision kiwifruit detection system is proposed. In order to improve the recognition accuracy, the R component map is used as the input map, and the median filtering method is used to denoise and extract the surface defect features of kiwifruit. On this basis, the least square method is used to establish the ellipse fitting curve to extract the size features, and the mean and standard deviation of H, I, S are used to extract the color features; At the same time, 30 models were selected to establish the deep extreme learning machine (DELM) model to identify kiwifruit defects. Finally, 326 kiwi images were constructed, and the results showed that the correct recognition rate of the depth limit learning machine model was 97.5%, which could achieve accurate grading.