O.I. Shelukhin, Yu.A. Ivanov, M.A. Smichok
The problem of video quality estimation is studied now both with theoretical and with practical points of view. New objective metrics of video signals quality evaluation, representing mathematical models imitating subjective appraisal are developed. The traditional way to assess the quality of digital video processing is the calculation of Peak Signal-To-Noise Ratio (PSNR).
Video transmission on wireless networks implies possible lack of synchronization between original sequence and the decoded copy on the receiving side as because of an unpredictable transmission medium data packets can undergo serious distortions or, moreover – to be lost. Therefore synchronization between analysable video sequences is necessary for carrying out manually. Thus, it is important to develop the software considering to the full these singularities and capable correctly calculate parameters of quality in case of miss of one or several frames in video sequence.
In this paper singularities of implementation of the developed software are stated, allowing to estimate quality of decoding of streaming video in the conditions of synchronization disturbance of transmitted and received video streams.
Main principle of the program is estimation of imported distortions in a video series taking into account frame comparing of the initial and accepted video sequence. For each frame of the accepted video sequence (i.e. transited through data communication networks) ratio PSNR is calculated.
It is shown that at asynchronous video sequences processing by means of the developed software «a splice» method gives the least errors. The greatest distinctions of the original and recovered distribution functions shows a method of an insertion of «zero» values. Thus with increasing of lost packets percent these distinctions essentially increases that does the given method unsuitable for practical application.