VP4000
VP4000
4K Picture Quality Analyzer
4K video real time double stimulus (a comparison method) quality analysis
The The frame delay of 2160p full sample 4K reference video and 4K test video are compensated to fit exactly to each other by pixel base, and a real time objective quality score (DSCQS) are calculated by the ITU-T J.144-based assessing method in conjunction with the differential values (PSNR). This scheme is best suited to the encoder evaluation, online/standby comparison, inspection geared to video servers, and so on. For conventional 1080i image evaluation, a single PROBE4000 board can serve the purpose.
Input video specifications
- Resolution/frame rate: 1920 x 1080, YC 4:2:2, 10 bit, 59.94p/50 (with one PROBE4000 board)
SDI: 3G-SDI Level A or Level B Quad input, SMPTE 424M/ SMPTE 425M
Image Partition: Square Division(Quarters)and 2-Sample Interleaves - Resolution/frame rate: 3840 x 2160, YC 4:2:2, 10, bit 59.94p/50 (with 4 PROBE4000 boards)
PSNR and Objective Quality Score
Since the objective quality score obtained from the double stimulus analysis is usable as a subjective score in its own given by a human eyes (QoE: Quality of Experience), whether or not user quality requirement is met can precisely be judged on a frame basis. In general, at a broadcasting station, for instance, degree of picture quality degradation not exceeding 12 % (4.5 or more in 5-point scale) is acceptable from the viewpoint of service. DSCQS : Double Stimulus Continuous Quality Scale.
ITU-T J.144 picture quality measurement algorithm
Core parameters for picture quality analysis are calculated complying with ITU-T J.144 standard.
PSNR and objective DSCQS (Double Stimulus Continuous Quality Scale)
Very locally occurred pixel error can be detected which PSNR is incapable
It is quite difficult by PSNR to detect a very locally occurred pixel error such as single block noise or a dot inversion because PSNR is an averaged value where a minor error becomes negligible small. However, VP4000 can identify a pixel error from compression noise by a pixel by pixel investigation algorithm.