Study On Identification Technology Of Wormhole Sunflower Seeds Based On Machine Vision

With the rapid development and gradually mature of the machine vision technology,visual detection technology is widely used in various fields. Because of the advantages likefast processing, abundant information and non-contact, the machine vision theory andapplication for quality inspection of agricultural products has become a hot topic ofautomatic detection and sorting of agricultural products. Based on theoretical study andexperimental analysis, this paper analyses and researches the related visual characteristics ofsunflower seeds. Through extracting characteristics of the surface texture, the edge and thewormhole, this paper realizes the detection and recognition of single wormhole sunflowerseeds and multiple wormhole sunflower seeds. The main contents are as follow:1. According to the characteristics of sunflower seed individual, this paper analyses thespecific requirements of sunflower seed image acquisition, completes the selection anddesign of the industrial camera, the camera lens, the light sources and the light box, andbuilds the visual detection system for wormhole sunflower seeds;2. Analyze the characteristics of the single sunflower seeds individual, and put forwardthe method of image acquisition for the single wormhole sunflower seed. According to theimage information, research and analyze the number of patches, white surface proportionand worm eaten characteristic of the single sunflower seed. Use the image thresholdsegmentation, connected domain identification,"light absorbing effect" and edge analysisalgorithm to realize the detection and extraction of the three kinds characteristics of thesingle wormhole sunflower seed, and apply support vector machine (SVM) to identify thewormhole sunflower seed accurately.3. According to the actual application requirements, analyze the characteristics and theacquisition method of the multiple wormhole sunflower seeds image, and research thewormhole characteristic in the face and the wormhole characteristic at the edge in themultiple wormhole sunflower seeds image. Use the "light absorbing effect" of a hole andpolygonal approximation algorithm to design the efficient identification method forwormhole sunflower seeds in practical application. And the accuracy of the method is95%.4. According to the features of the multi wormhole sunflower seeds identification andsunflower seeds processing, design the wormhole sunflower seeds sorting system based onmachine vision. Analyze and elaborate the function and design method of the conveyingsystem, the visual identification system, the sorting system and the turning system in detail,and build the experimental device preliminarily.