Research On Inspection Technology Of Dehydrated Garlic Slice Based On Computer Vision

Automation of agriculture producing is the future trend by computer vision inspects the quality of primary products. There has made much research on the quality of primary products. Select three grades dehydrated garlic slice and research the identification of dehydrated garlic slice by computer vision. On the basis of pattern recognition analyze image and extract feature parameters, which reflect the morphology feature of dehydrated garlic slice. In the research using the neural network recognizes and grades dehydrated garlic slices.First, a computer vision system was developed to detect the dehydrated garlic slice surface feature and grading the dehydrated garlic slice.Second, algorithms of image pretreatment were studied. Comparing the background segmentation results of dehydrated garlic slice images in R, G, B channels by iteration and OSTU, background segmentation of color image was proposed, which applied OSTU to image R channel, and the dehydrated garlic slice region was marked by scan line seed fill algorithm, and then get the dehydrated garlic slice images.The morphological structure of dehydrated garlic slice is analyzed. Two group parameters including color and shape show the morphological structure of dehydrated garlic slice. Color parameters were denoted by HSI color model.Then network was trained by LMBP algorithm. Through analyzing the influence of feature parameters on recognition rate, four parameters which could reflect the shape features perfectly were selected. The experiments results showed that 4-16-1 network structure had higher recognition rate, which could reach 85%.Finally, A dehydrated garlic slice grading system for dehydrated garlic slice surface features was designed with VC++ 6.0. The software has friend interface and stronger function, which can finish background segment and smooth of the image and extract dehydrated garlic slice and figure out the feature parameters.