The Study On Comprehensive Algorithm Of Calibration Based On Image Measurement

As a state-of-the-image measurement technique based on computer vision, visualmeasurement technique, with many aspects of the advantages, such as high efficiency,low cost, high accuracy, has a very wide range of applications. Binocular stereovision is an important branch of computer vision. Binocular measurement withreal-time measurement and three-dimensional measurement accuracy, spends the costto be much smaller than other methods, therefore having a wide range of applications.In addition to the measurement of the actual objects, it also has achieved rapiddevelopment in other areas.As a three-dimensional measurement techniques using car body to be standard,this paper focuses on the goal of improving the accuracy of the camera, based oncamera calibration process, putting emphasis upon this critical issue that how toimprove the calibration accuracy. Binocular camera calibration system uses adual-camera combined with the four calibration plate, each with a calibration disc oneach wheel tire. The domestic measurements existed are that each binocular cameraaligned with one calibration target, while the design of this paper is each binocularcamera aligned with four calibration target. The measurement technology thatcalibration target by a fixed rigid body in the middle, creates a revolutionary way ofmeasuring between the calibration object under occlusion calibration. In the binocularcamera’s calibrating process, the outside noise greater impacts on the grid edge blurand distortion owing to infrared light, which will inevitably cause the coordinates ofthe corner extraction errors. Compared to images edge, the centre has less impact ofoutside noise. Therefore, this paper improves Zhang Zhengyou’s calibration methodwith the grid replaced by round extract center coordinates of the calibration andapplication in image acquisition with a poor shadow method, effectively eliminatingof the effects of image background interference.Based on the mathematical model of the wheel alignment parameters, we canuse the camera capture sequence of images of the circular target plate with the wheel movement and circular pieces of sequence images taken from different anglesthrough the wheel movement to the feature point detection, and then by calculatingthe relationship between image coordinates of the feature points and worldcoordinates to do camera calibration. The calibration results will be as the initialvalue for the improvement of the nonlinear optimization algorithm, using improvedparticle swarm optimization algorithm to optimize the camera parameters, taking intoaccount factors such as population size, selecting the appropriate inertia weight andlearning factors parameters, to achieve better convergence.This paper firstly considers how to calibrating object which has a depth andcalibrating frontways target and rearways target synchronously in a calibrationprocess. The calibration process has this issue of how to focus in-depth study. Thepaper also proposes idea of partitioning the image for the first time, which discussesin depth information of the case of multi-target calibration, by adopting optimizationalgorithm to get the best image distance in which we can make clear calibration offrontways target and rearways target, and the resulting calibration parameters takinginto account both calibration target frontways and rearways. The experimental resultsshow that by using a nonlinear optimization algorithm to optimize the results ofcalibration, higher precision calibration parameters can be achieved, and the design isapplicable to calibration of block material scene between the objects, by finding thebest image distance this paper considering the multi-target calibration of the depth ofinformation, therefore has a good value.