Аpplication of wavelet transformations to increase the efficiency of the information system of forensic expertise

Abstract

An analysis of the functioning of the forensic information system is carried out. It is determined that the huge information capacity of images in one way or another limits the possibilities of their use in solving various scientific and practical problems, including in forensic science. As a result of the analysis it is established that the rapid progress of technical means of photography, video recording, telecommunication technologies expands the possibilities of traditional means of capturing photo, video and new (mobile, network, specialized, space) with new data formats, requires constant updating of special expertise. in the field of digital photography. Existing methods of image processing, however, do not fully solve the problem of their effective representation, which makes the search for new effective methods of image representation relevant. To solve this problem, the use of orthogonal transformations is proposed. The article solves the current scientific and technical problem of analysis of various adaptive orthogonal transformations of information to identify more effective. To solve this complex and multifaceted problem, the article studies the so-called wavelet packets, or adaptation in the frequency domain; the algorithm of a double tree, or adaptation of the basis of decomposition both in frequency, and in spatial areas is analyzed; researches of dimension of library of bases for all transformations and their computational complexity are carried out. As a result of the conducted researches it is established that prospects of application of this or that algorithm depend on the concrete application. In addition, it is likely that better results can be achieved if we separate the segmentation process from the transformation using wavelet packets. Currently, effective segmentation algorithms have been developed that can be successfully applied. After segmentation, each segment is reduced to a rectangular shape, and it is transformed using wavelet packets. Thus, the main goal of the study is solved - it is established that adaptive orthogonal transformations, in comparison with traditional wavelet transforms, allow to improve up to 10% of the value of the peak signal-to-noise ratio at low coding speeds.