Research On Friut Surface Quality Detection Based On Digital Image Processing

China is a fruit production country in the world whose annual production value occupies first place of the world. However, it is not a fruit trading power for its annual fruit export volume was only3%of total production value. The main reason for our lack of international competition in fruit is the backward commercialization level after harvests. The fruit post-harvest in China is simply classified according to the size and weight, regardless of classification standard and timeliness; while the fruit post-harvest overseas have been processed by a series of steps:selecting, grading, washing, waxing, ripening, packaging, and cooling. Thus, it is very essential for our country to improve the fruit’ commercialization level after being harvested and fruit quality check is the most important part in the fruit commercialization.The fruit appearance is an important indicator of the quality check. In this paper, combining digital image processing knowledge with the related experiment helps study the method of fruit surface quality check.First of all, in order to retain more image information in the preprocessing link of the fruit image (the color information mainly focused), the method that is suitable for the color image preprocess of the fruit images is analyzed in the paper.In order to achieve color category classification more accurate and close to the human perception in the fruit surface color detection, the method of the fruit surface color detection in the HSI color space is analyzed in the paper. First of all, the Hue rotation method is presented with regard to the red singularity problem of the HSI color space. The experiment, by rotating the color plane anticlockwise of120degrees, proved it has avoided the red singularity problem effectively. Then a histogram analysis of H color component through the color rotation is carried out, to obtain the color range of different types of fruit. Finally, it is classified by using the minimum distance classifier. The experiments show that the classification can be achieved according to the fruit surface color.At present, the research on fruit surface defects has mainly focused on defect detection, and the surface defect classification is seldom studied. For the fruit surface defect detection, a new detection method based on attention selection mechanism is presented in the paper. In this way defect region can be segmented from the fruit images by means of this method. The experiments show that the defect region partitioned by using this method is much closer to the judgment of the fruit surface defects. Then, based on detecting defection of the fruit surface accurately, the integrated feature of texture and color in the region of defection is extracted, and the defect is classified by support vector machine classifiers. Finally, a better classification result can be got from the experiments and the fruit surface defect classification can be achieved.Finally, the research achievements and deficiency are summarized in the paper, and the idea and prospects to further complete the detection technology for fruit surface quality are propose.