Shape Of Image Recognition Technology And Applied Reseach

Based on machine vision system of industrial product defects detection, had been widely used.The shape of the image detection problem is a class of problems frequently encountered by the defect detection of industrial products. For such problems, first you want to extract the shape of the image of the target, this step involves the key technologies of image preprocessing and segmentation; The second step, Shape analysis of the shape of the image segment, to make it better able to distinguish it from other objects, so as to achieve a higher recognition rate, this step involves key issues such as shape analysis and recognition.This paper pointer type instrument dial segmentation and badminton craft ball defects detection actual requirements as the background, study the intensity of illumination condition of uneven distribution of image segmentation technology, and give the actual badminton can be applied to industrial production based on shape recognition of the defect inspection of algorithm and software and hardware platform. The main research results and innovation points summarized as follows:1, Pointer type instrument image preprocessing and divisionIn the process of verification, because light condition and the limits of the focal length of the lens, dial reflection in the glass and the movement of the pointer, lead to obtain image region of uneven distribution of light and image slants dark. In order to achieve in this condition of target area segmentation, first the Laplace sharpening and Retinex multi-scale enhanced key method, this method can not only enhance contrast, but also increase the overall image intensity, these processing is very favorable for the back of the image segmentation. With the dynamic segmentation Otsu after pretreatment algorithm of image segmentation. Through the experiment, this method to uneven distribution of intensity of illumination, the image of the whole partial dark image segmentation with good results. Because the collected badminton craft ball image also has this problem, make this method can also very good application to the badminton craft ball shape the image extraction.2, Based on VC++badminton craft ball defects detection platform This system includes industrial camera, light, the light controller, the agency placed and conveyor belt device, software platform for VC++. Software system including image acquisition, image processing algorithm, communication debugging, size calibration and camera debugging module. This system the image acquisition and image processing module separated, so as to realize the industrial camera when change not need to the whole system of big changes, to avoid the industrial camera acquisition module of repeated development, and image processing module integration many common image processing algorithm, can through the choice of different acquisition module and the combination of different image processing algorithm for different object to identify the experiment.3, Badminton craft ball defect detection algorithmIn badminton craft ball in practical production, there are more than ten defects, all by artificial visual. On the analysis of the types and a large sample defect defects after investigation, Sure95%of defects types are by16feathers leaves caused by uneven distribution, in the image for feather leaves mutual reflect of the formation of the16of the shape of the regional gap uneven distribution. Therefore, badminton craft ball defects detection is the main content of16the distribution of regional gap even tested. This paper used a radius constant torque feature extraction method, this method is not sensitive to the symmetry, no matter for boundary, or region has good recognition effect. The final selection of five very shape image radius constant torque characteristics, the ball in vitro round central degrees, feather leaves16vertex and ball head average distance circle, and circle deviation. a total of eight characteristic parameters of vector image features, then the euclidean distance by the recognition. The experiments show that the method can effectively used in badminton process defect inspection of the ball.