Research Of Calibration Of Special Sensing Range Cameras

Accurate camera parameters are the basis of most computer vision techniques.The state-of-art camera calibration technology can calibrate most cameras,but there is no universal calibration method for the calibration of some special sensing range cameras.The Zhang Zhengyou method uses the translation matrix of different calibration plate positions when building equations.When a camera with a special sensing range is calibrated using the Zhang Zhengyou method,it always takes a defocused picture and cannot obtain effective information.The calibration results are usually poor or even impossible to calibrate.For this issue,the work and conclusions made in this paper are as follows.In the first two chapters,we briefly summarized current research results and hot issues of camera calibration and proposed the deficiencies of existing methods.In the third chapter,based on the related knowledge of multi-view geometry,a calibration method based on active vision for macro camera is established,and the experimental verification is completed.This method uses a translation vector between different calibration plate positions for calibration,resulting in good calibration results.This article also designed a simple method,eliminating the need for a mobile platform.The method utilizes the high flatness of the liquid crystal and the neat alignment of the liquid crystal to project the calibration object onto the liquid crystal panel.The change of the position of the calibration object is equivalent to the translational movement of the camera,which greatly saves the cost of the calibration equipment.Experimental measurements,the re-projection error after calibration within 0.2456 pixelsIn the fourth chapter of this paper,two kinds of calibration methods under the condition of camera defocus are proposed.Using the property of sinusoidal structured light correlation,the first method performs phase encoding on common calibration objects.In this method,the structured light is used to encode the calibration object,and an accurate focus camera is used to record the phase information of each feature point.The out-of-focus camera performs phase decoding based on the captured image of the blurred structure light,and then obtains the exact image coordinates of the feature points of the calibration object.After calibration,the maximum deviation of the focal length from the true value is within 0.5%.The maximum pixel reprojection error is 0.17 pixels.In the second method,the method performs phase encoding on the liquid crystal display panel instead of the normal calibration object.The phenomenon of gamma conversion is common in liquid crystal display panels.This section discusses methods for suppressing gamma transformations using sinusoidal fringes using the Floyd-Steinberg Dithering algorithm.In this section,the practicality of the Floyd-Steinberg Dithering algorithm is demonstrated through simulation experiments.During the use of the method,the structured light is used to encode the liquid crystal panel,phase decoding is performed according to the captured image of the blurred liquid crystal panel,and then the accurate image coordinates of the feature points are obtained.After calibration,the deviation of the focal length from the ideal value is within 0.68%.The maximum pixel reprojection error is 0.17 pixels.After experimental verification,both methods can achieve satisfactory calibration results.