Image Processing And Image Recognition Based On The Robot-soccer

With the wider and wider use of machine vision, people's research on image processing has become more and more exigent. The vision subsystem of robot-soccer system is a better platform for the research of image processing, whose requirements of real-time, veracity, and adjustability are very suitable for the modern industrial production. The study of the robot-soccer vision subsystem must promote the fast development of machine vision.Image Partition and image recognition are two important parts of image processing. Image partition technology has developed for many decades, but it is short of a uniform partition theory. Lots of algorithms are only valid to numerous and complicated specific questions. Color image possesses the multidimensional characteristic in essence, which makes a great deal of effective image partition methods in grey images not suitable any longer. Color image partition and multi-object real-time recognition have become one of the highlights and difficulties in the research of vision system. Improving the adjustability of the vision system to the illumination change of surroundings will play a very important role of developing the application field of vision system.This paper works over the following contents: the theory and method of color image partition, the idea of local transition, and the high intelligent self-correct means of harmonizing vision and action, and so on. The main contributions are generalized as follow:1. In the theory, firstly, depending on the set theory, offers the new description of color image in the form of the set function;secondly, extending the concept of eigenfunction, defines the concept of the multi-value eigenfunction and brings forward the new thought of dividing the image coordinate set and further achieving the image partition based on multi-value eigenfunction;thirdly, gives the math description of achieving image partition based on the multi-value eigenfunction, fourthly, presents the idea of the local color space transition and local pixels correcting aberrance.2. In the technology, gives the color image partition method of Sample Threshold Value, which is a kind of characteristic clustering method. Through analyzing the samples in the color space, builds the object sample's superset in color space, describes the result of the multi-object clustering with the eigenfunction in color space, and finally, stores the result of the eigenfunction into the object characteristic repository. In the process of the real-time image partition, using the eigenfunction to realize the partition of image coordinate set is very convenient and easy.3. Muti-target visual tracing. To improve visual tracing precision and environment adaptability, it presents a method of robot-soccer visual tracing based on action-vision coordination. The method validates the coherence of target action and command by vision. It can keep uninterrupted tracing in visual blind area, and correct the target tracing error.