Aiming to achieve the automatic detection and accurate identification of pebrine images, the fuzzy contrast enhancement algorithm was utilized to enhance the contrast of the target image in order to improve the image’s quality; Owing to the color character of light green for the pebrine, the image segmentation technique based on the HSI model can be applied to extract the pebeine image, at the same time, the morphology theory can be adopted to remove the noises such as the hole noise and point noise, and to separate the bond particles; The region labeling can be done on the binary image after the image segmentation, then the shape parameters of pebrine can be extracted, by making full use of the feature parameters, the method of neural network based on genetic algorithm is applied to recognize the pebrine image. These experiments show that the method has achieved satisfactory image recognition results.