Solving the problem transform of size images on series base points found on frames in the problem of stitching images

Stitching several images into a single content is an important task for many areas, including video analytics, machine vision, security, etc. The operation of combining is based on the solution of a number of operations related to the allocation of local features, the selection of transformation matrices of objects, changing the scale and saturation of the image. The process fixing frames is he’s with a number of difficulties, includes the appearance of a noise component in the image, blurring, various non-linear effects of the optical system, as well as problems associated with incorrect operation of data preprocessing software. To select the key points used as alignment points is a complex problem, the solution of which is the subject of many works. However, the transformation of coordinate systems of images with combination frames is a difficult problem, since the system is nonlinear. The article discusses the solution to the problem of changing the position of points in clusters, under conditions of their nonlinearity. The method and algorithm that implements the procedure of nonlinear transformations size over the entire image is presented. Operations include vector analysis, as well as the results of changing the scale of a group of pixels. The set of test data shows the possibility of applying these approaches to change one of the images according to the specified combination points by the operators.