Research On Deformity Correction Algorithm Of Triangular Block Image Based On Spatial Mapping

With the rapid development of artificial intelligence technology,all walks of life have higher and higher requirements for the ability of computer vision.Computer vision is a series of processes such as target recognition,matching,and further processing of the image,so that the image after computer processing becomes more suitable for human eyes to observe or transmit to the instrument detection of the image.In the process of image shooting,the original image will have geometric deformation problems in geometric position,shape,size and orientation due to shooting Angle or capture,which brings great challenges to subsequent image processing processes such as image fusion,image denoising,image recognition and image edge detection.Therefore,solving the problem of geometric deformation in the process of image shooting can further improve the processing ability of the computer for the subsequent process of image,and increase the accuracy of subsequent image matching and recognition.In view of the situation,this thesis proposes an adaptive image distortion correction algorithm based on triangle segmentation for quadrilateral target images.The main contents of this thesis are as follows:1.Image preprocessing.Pre-processing distorted images.Firstly,adaptive filtering is used to remove the shadow or other noise of the image,and edge detection is used to rough locate the edge of the image,and then mathematical morphology is used to expand the edge contour of the rough location of the image.Finally,the maximum connected domain of the image edge contour is calculated and extracted to filter the internal information of the image and retain the maximum edge contour of the image.2.Image vertex positioning.The image edge discrete points were found,and the data information of the image to be corrected was introduced into the spatial geometric coordinate system.In the spatial geometric coordinate system,the edge of the image was accurately judged and the feature vertex of the image was found by filtering the slope variance of two adjacent points of the discrete edge points of the maximum edge contour of the image,and the method of linear fitting and gradient descent was integrated.3.Three-dimensional homologous mapping of image.The quadrilateral target image is divided into two triangles according to the diagonal,and ensure that each triangle image of quadrilateral target image can be corrected accurately.Firstly,the feasibility of triangular object image correction is verified,and the triangular object image is taken as an example.Then,the mapping relationship between the two segmented triangle images and the standard right triangle images is calculated respectively,and the two segmented triangle distorted images are mapped to the standard right triangle images by using the homography transformation of three-dimensional space.Finally,the two corrected right triangle images were combined,and the reasons for the failure of the combination of corrected images were analyzed in the process of combination.The mapping relationship was repositioned through mathematical formula derivation,and the two right triangle images were recombined according to the new mapping relationship,completing the transformation from irregular image to regular image.A large number of experiments show that the proposed algorithm is superior to Hough transform,Radon transform and traditional mapping algorithm,and solves the limitation of the traditional algorithm in correcting quadrilateral image,improves the correction effect of distorted image,solves the geometric deformation problem of polygon image well,and provides a new technical method for image processing of distorted image.