Research On A Model Of Camera Calibration Based On Binocular Stereo Vision

vision technology is one of the hot topics in computer field recently. Vision mea-surement technique is an advanced non-contact technique that has the advantages suchas high speed, high accuracy and low cost. Therefore, it has got wide development andapplication in industry, especially in manufacturing industry for testing products. Thebinocular stereo vision is a method that observes the same scenery between two di?erentangle to get the pictures of the di?erent visual angle. and it can get the three dimensionalinformation of the environment by using the knowledge of geometry theory and calcu-lating pixel warp on two images. In order to research the model of camera calibrationbased on the binocular stereo vision, the main work of this thesis is as follows:1. Not considering the e?ects of the aberrance of lens, we have built a linear imag-ing model based on the pin-hole camera model. and we include the transform rela-tionship between pixel coordinate system and other coordinate systems with geometryknowledge. Then, we find out a method to transform object point coordinates of worldcoordinate system into pixel coordinates of pixel coordinate system.2. We included two methods to solve the parameters of model we have built in thisthesis, one is based on the second problem which gave out from the model we shoulddiscuss, the other is to use two-steps method to solve the parameters of the model wediscussed.3. We provide the definition of precision and two kinds of stability of the modelproblem. According to the conclusion of geometry, we can find the geometric centerof ellipse in pixel coordinate system by using the method of centroid measure. So, thequantitative analysis of precision can be achieved; In the analysis of stability of themodel problem, we verified our algorithm of camera calibration with the definition wehave involved in previous.4. By using the knowledge about the binocular stereo vision, we have designed amethod to find the relative position of two cameras, and also gave out the expressionformula to calculate the relative distance between two cameras.