A method of quasi-circular objects detection

In pattern recognition and computer vision, quasi-circular detection and parameters determination are important tasks. In view of the special characteristics of quasi-circular objects, firstly, the color image is processed with the following pretreatment methods: image grayness, image smoothness, image binaryzation and so on, and the high quality binary image can be obtained; then the distance image of the binary image is achieved by using distance transform based on boundary-stripped algorithm; in the distance image, the quasi-circular objects are detected by searching and flattening the maximum pixel areas one by one; so far, the end results are completed. Simulation results show that this algorithm has good effects, and it can detect and calculate the quasi-circular objects precisely and rapidly.