Research On The Shape Identification Of Orange Based On Computer Vision Technology

The orange industry is one of the mainstays of rural economy in the major agricultural producing areas of Southern China. At present, the orange grading depends on manpower and machines, which cause slow efficiency.Appliing of computer vision technology, in automatic orange grading can increase the commodity value and productivity. Referring to relative studying on computer vision, this paper introduces an automatic grading system and summarizes a set of arithmetic of online automatic grading according to the size, color and shape of orange with the assistance of computer vision technology. Furthermore, image pretreatment method and the estimative method of grading pattern are presented. Advanced and applicable OTSU method of segmentation is applied to implement background segmentation. Median filter and mathematical morphologic operator are used to eliminate noise. Projective area and picture hue describes the size and color of the orange respectively, which will also be applied to identify the disfigurement of orange. With simple arithmetic being used in identifying the size, color and shape of oranges, the grading speed could be increased.The technology of online automatic orange grading according to computer vision has been successfully implemented with the application of Visual C++6.0, MFC and other class libraries based on Windows2000 system. The benchmark and level of grading can be adjusted according to the request of client and market. Through some trial applications, the grading system has been proved to be able to meet every requirement of the precision inspection of factories and has reached the anticipated objective.