In order to detect the broken shiitake, a automatic detection system of shiitake was developed based on machine vision. Identification algorithms based on curve evolution and shiitake edge grayscale were presented. First, the background of shiitake images was removed, the edge of shiitake was tracked and then the coordinates of shiitake boundary was obtained. A closed curve was composed of these coordinates. Two initial curves could be generated from the interior and exterior of the closed curve respectively. Two final curves were evolved on those two initial curves, which met the condition of specific termination criterion. Two parameters (Nin，Nout) were extracted from the difference of final curves and initial curve. These parameters could determine the broken extent of shiitake and shiitake edge region were sampled with the method of morphology. Then four parameters could be extracted from the sequence of the gray scale of sampled regions .These parameters were mean (μ), variance (ρ), average width of peaks (Lp) and the maximum width of peaks (Lmax) respectively. The four parameters were analyzed with the method of pattern recognition and the broken extent of shiitake was obtained from the processing results. The final result was given with the results of both curve evolution and grayscale analysis. Experiments show that the accuracy of final shiitake detection model reached up to 88.33% , and the selection speed was 98 per minute.