Other Works in Objective Video Quality Measures

In [153], ITS, MPQM and NVFM quality assessment metrics are evaluated for a wide range variation of bit rates. The study showed that the metrics based on HVS give better results than those based on pure mathematical calculations like PSNR. However, the study showed that MPQM and NVFM give unreliable results for low bit rates. In [104], a methodology for video quality assessment using objective parameters based on image segmentation is presented. Natural scenes are segmented into plane, edge and texture regions, and a set of objective parameters are assigned to each of these contexts. A perception-based model that predicts subjective ratings is defined by computing the relationship between objective measures and the results of subjective assessment tests, applied to a set of natural scenes and MPEG-2 video codecs. A logistic curve is used to approximate the relationship between each objective parameter and the subjective impairment level. Thus an impairment level is estimated for each of the objective parameters. The final result is computed by a linear combination of the estimated impairment levels, where the weight of each impairment level is proportional to its statistical reliability. Another objective video quality assessment method, which is similar to that presented in [104], is given in [35]. In the article, a morphological video segmentation is used to estimate the subjective quality. Each frame is segmented into three regions: homogeneous, border and texture. The segmentation process uses a collection of morphological tools such as connected smoothing filters, morphological gradients, and watershed. This method has the same characteristics of that presented in [104]. However, the objective results do not correlate with subjective ones, as the MSE ranges from 24.5 to 67.3 % for the 5 tested video sequences. In addition, the only tested impairments are those resulting from encoding/decoding, and no other kind of impairments are studied. In [130,131], an objective picture-quality measurement model for MPEG video is presented. The theory of this method is to design a picture-quality model for each kind of known distortion, and to combine the results from the models according to the perceptual impact of each type of impairment. The model consists of quantifying both the blurring and the blockiness effects resulting from the encoding/decoding processes. The blockiness is estimated by using a blockiness detector, and the blurring is measured by a perceptual model. The outputs are then combined to give the perceptibility of the distortion. The final rating is either the output of the blurring model or the blockiness detector. The selection is based on the average amount of blockiness measured in the decoded sequence. When it is high, the blockiness detector's score is selected. Otherwise, the score of the perceptual model is chosen. In [62], another objective quality measure is based on the time transition of quality of each frame. By obtaining the quality of each frame and applying a weighting function of all the frames in the sequence, the quality is obtained for the whole sequence. The employed frame-quality assessment method is similar to that in [130,131]. In [33], a study of video quality metrics for MPEG-2 video is given. The authors are more interested in end users' points of view for packet video. A state-of-the-art in the definition of video quality metrics is given. They evaluate also ITS and MPQM metrics versus simply the bit rate as as the work presented in [153]. From that the ITS gives very bad results for low bit rates. In [119], an objective video quality assessment method for videoconferencing and videophone systems is introduced. It is based on the Temporal Frequency Response measurement technique. The authors use only four test conditions to evaluate the performance of their method. In addition, the only parameter investigated is the bit rate. From [143], it is clear that the correlation is not always good, and some times there is no correlation at all (the same as the results obtained from [24,130,131]). A state-of-the-art for some objective video quality assessment methods for MPEG-2 video is given in [102].