| Since the competition becomes more fierce, instant noodles' profits is shape decline. After entered into WTO, domestic instant noodles enterprises confront with more severe challenge. Compared with foreign product lines, Chinese product lines have raletive low automatic level. Unqualified instant noodles are eliminated by people, which resultes in great workload, low accuracy and bad food sanitation. As the technology of computer vision becomes more and more mature, it is possible that people are replaced by computer vision. The technology of computer vision will be a promise application in eliminated unqualifies insistant noodles.Recognizing the unqualified insistant noodles based on computer vision were discussed in this paper. The main study results of thesis were as follows:1．Machine vision system fitting for instant noodles product lines was made, and acquired dynamic imaging in our simulated product line.2．The image was preprocessed. According to the character of dough, around shock method was proposed to segment the image.3．To judge irregular shape of dough, The burr of edge was get rid of after "cut processing", then the area ratio of dough and exterior rectangle was available to identify dough rapidly. This was a simple algorithm and a shortcut.4．To judge bad texture of dough, Run-length matrix was simplified and scanning time and algorithm time were induced extremely.5．To judge deep-frying status of dough, the ratio of browning area anddough(colorratio), color-difference of dough(σ) and frequency of S(A) were available. BP neural network model was established and the total discriminant accuracy was93．3%. |
