Forensic system for surveilance camera image enhancement

V diplomskem delu je obravnavan problem izboljšave slik videonadzornih kamer v namene forenzičnih raziskav. V forenziki se tehniki pogosto srečujejo s problemom premajhne ločljivosti oziroma prevelike oddaljenosti nekega ključnega objekta na videoposnetku. V delu so predstavljeni obstoječa programska oprema za izboljšavo slik ter osnovni filtri in tehnike izboljšav, ki so lahko v pomoč. Diplomsko delo se nato osredotoči na bolj zapleteno metodo superresolucije. Problem se je skušal rešiti z algoritmi superresolucije, bolj natančno z bikubičnim in bilinearnim interpolacijskim pristopom. Proučevano je bilo, koliko superresolucija lahko pomaga v različnih primerih.In this thesis the problem of enhancing images of video surveillance cameras for purposes of forensic investigation will be addressed. In forensics technicians often come across the problem of low resolution, or the key subject being too far away on the video or image. First the software that allows technicians to process images and the basic filters and processing techniques will be presented. After that the thesis will focus on a more complex method called super-resolution. The problem of low resolution will try to be solved with super-resolution algorithms, more specifically using bicubic interpolation approach and bilinear interpolation approach. The main point of this thesis, is to learn how much the super-resolution can help in different scenarios