Research And Implementation Of Automatic Sorting System For Sunflower Seeds

From the 60’s of last century, the technology of image recognition and analysis is applied to medical examination, industrial processing and production. In recent years, the application of digital image processing and recognition technology is gradually coming into our daily life, such as rice grading, fruit and vegetable grading, and so on. Compared with manual identification, the automatic sorting system based on vision technology is not only fast, accurate, but also can be repeated. So far, the sorting of the insect and sunflower seeds depends on the manual operation. According to the market demand for sunflower seed production and improve the quality of products, the automatic sorting system of sunflower seed and its key technology has been researched, and automatic sorting equipment based on image detection technology of sunflower seed has been designed and developed. The main research contents of this paper are as follows:1. Based on the requirement of the system’s efficiency of sorting and the characteristics of the defect of sunflower seeds, the automatic sorting system of sunflower seeds is set up. System hardware structure, which includes the camera detection system, mechanical structure and power distribution box are designed.2. Target location is a prerequisite for the sorting of sunflower seeds. In this paper, we study the common localization methods, A coarse positioning method is proposed, which combines edge detection, region filling and object shape feature. Based on the characteristic of noise sensitivity of Sobel operator, the edge detection of sub sample image is carried out, and then the binary image is processed. To remove the background of small independent noise region, the regional denoising method is carried out; Throughout the region filling, we get target area without holes; Then extract the boundary and combine the feature of the target to locate the sunflower seeds.3. Target detection is the key to the sorting of sunflower seeds. In this paper, we improve the traditional Canny algorithm in the edge detection, and get more effective edge. The popular edge detection method is studied, and an improved IVC model and a LSPF model are proposed. After comparing the three algorithms, the improved Canny algorithm can get more efficient edge, and the LSPF model can obtain the effective target profile, and the performance of the improved IVC model is between the two.4. Target recognition is the guarantee of the sorting of sunflower seeds. Firstly, according to the characteristics of the defects of sunflower seeds, we can extract 7 kinds of effective features, then the common recognition algorithms are analyzed, and a classification algorithm of sunflower seeds based on Adaboost is proposed. In this paper, the Adaboost algorithm is used to train a large number of samples, and the training of the classifier is cascaded. Experimental results show that it can get very good classification results.5. In this paper, we designed and developed a sunflower seed real-time sorting software system based on MFC single document framework, and use opencv library, database technology ADO, serial programming, network programming technology and industrial camera driver library technology, on the windows platform. The system has friendly software interface, provides four major functions, including different permissions login function, positioning detection and identification of the debugging features, valve island parameters debugging features and binocular real-time detection function.6. The research work of this paper is summarized, and the future work is prospected.