Recognition Of The Numeric Identifier In The Study Of Animal Behavior

Based on image processing and pattern recognition methods to solve some specific areas of expertise related problem is one of the current applied research in the field of computer vision. On the background of studying the activities habits of bees,The main research contains the preprocessing of the digital identifier,the location of the circular area,the digital segmentation in the location,digital calibration and digital identification and other aspects.In the preprocessing stage of the digital identifier, through the gray of the original image, histogram equalization, spatial filter and a series of operations, achieve the purpose of eliminating the irrelevant information in the image and restore the true useful information, further improve the image segmentation and the reliability of correct identification.In the edge detection stage, through several edge detection operator for image preprocessing to extract the edge well and make a comparative analysis of results. the results showed that the Canny edge detection operator were even similar the visual effects. Using the improved methods and regional positioning Hough circle algorithm combined with the theory circular sub-pixel location algorithm position the circular area, an experiment showed the good results.Automatically calculated by Otsu threshold segmentation method to achieve the partition of digitals in the identifier region, and the tilt correction of the single digital after the segmentation, calibration using a recursive way by gradually binary comparison the ratio of image width to estimate the size of angle.The identification phase of the digital identifier, through improved Hausdorff distance template matching and support vector machine multi-classification complete the identification. By comparing the experimental results of OAO, OAR, HAH, the results shows that HAH better meet the needs of this paper.