Detection images with similar content

The article discusses methods for detecting images with similar content in a search collection. The purpose of the work is to find the optimal solution (algorithm) for visual search. This search is complicated by the difference between the target image and the user's request. Another difficulty is the presence of invariant distortions, for example, image rotation, changing the scale and angle of view, as well as searching for an image component. The result is a list of images in a certain sequence with the same content for the query and their location. Scientific novelty lies in the use of convolution and subsampling operations to search for similar images in the desired image, in the development of an algorithm for more accurate search in case of image modification. Pairwise comparison of images is carried out using a neural network approach.