Objective and subjective image quality evaluation for security technology

The paper is devoted to the impacts of image compression algorithms on security image data. It compares three fundamentally different evaluation techniques of image objective criteria, subjective criteria and identification. We have selected two typical security image data (a car plate and a face) with different initial quality and we applied three different compression techniques-two professional (JPEG and LuRaWave-LWF) and one implemented (Karhunen-Loeve transform, KLT). A set of compressed images differing in compression rate was derived from each original image data. Finally the MSE as an objective criterion, the subjective image quality according to the ITU-R Rec. 500 and the identification measure were evaluated and compared.