Study On Detecting Defect Of Appearance For Northwestern Melon

Gansu Province is a main producing areas of melon, including white melon, the old and new varieties are about 20 kinds, in 2007, more than 400,000 tons melon production, however, the economic benefits is not very prominent, which is one important reason is to ignore post-harvest processing. currently, gansu melon is no classification and direct sales, quality is no specifications, so the economic value and competitiveness greatly reduced.The superiority of machine vision classification is a comprehensive inspection in in the size, shape, color, defect maturity, etc. the works will be freed from the heavy monotonous work.different results will be avoid of the same person at different times.this research will use machine vision technology defect detection of jinhongbao one of the northwest series features varieties of melon.In this research, the difficulty is that it is easily to confuse between muskmelon's defect and stem and calyx.in order to reduce the error rate the of muskmelon's defect detection,so we construct an automatic defect detection system based on support vector machine.four textural parameters features and twelve color features of combinations from RGB are tested. through the experiments, two textural and four color features with good discriminability are selected and treated as the complex features, the results indicate that with the complex features and support vector machine, the error rate of classification on the muskmelons was only up to 7.8%.