The Research Of Non-reference IPTV Quality Assessment Model

Currently, one of the most widely used Non-reference IPTV quality assessment model is MPQM. The principle of MPQM is described in detail. Two disadvantages are proposed:first, MPQM assess IPTV video quality through the network packet loss simply, it can not assess content of the video image; second, even in the same network packet loss rate, the video quality will be different because the different importance of TS packets which are loss, MPQM do not consider the impact of video quality caused by the TS parameters error, can not assess the quantity of the packets which can be decoded correctly. To solve the two disadvantages, a macroblock-based non-reference assessment model MBNM and a TS parameters-based non-reference model TSNM are proposed.According to solve the first disadvantage of MPQM model, the macroblock-based non-reference assessment model MBNM is proposed. The main technical background of this model is MPEG-2, which is an international standard developed by the Moving Picture Experts Group (MPEG). The MPEG-2 defines three image formats: intra coded frame (I-Frame), predictive coded (P-Frame) and bidirectional coded frame (B-Frame). There are six levels in the coding structure of MPEG-2, each level plays a different role.Since the human eye can not recognize the single-point noise of image, only detect massive damage, MBNM model is proposed from the macroblock layer., which not only considers the loss of macroblock, but also analyzes the injury caused by the loss macroblock quantitatively both in space-domain and time-domain, reflects the real quality of image accurately.In the process of proposing MBNM, the concepts related with the macroblock are defined firstly, such as the row and the number of macroblocks in a row. Calculations that how to get the total number of macroblocks in a image and macoblock coordinate are given. Then, according to the prediction method of the three frame formats, space-domain diffusion factor and time-domain diffusion factor which can analyze the influence of the loss macroblock both in space domain and time domain are proposed by analyzing the prediction method of MPEG-2 in detail. According to frame prediction method, the macroblock in I-Frame is only affected by the space-domain diffusion factor, but affected by both factors in other kinds of frames. Based on this, we can get the injury status of each kind of frame quantitatively, so that we can get the total injury quantity of macroblocks in a video frame image. Based on actual observation, the quality of video frame image is inversely proportional to the quantity of injury macroblocks. Thus, Non-reference assessment model MBNM is proposed which can be used to measure video image quality through analyzes the number of injury macroblocks quantitatively, and the assessment process of this model is designed. The assessment result has five levels, correspond to different video quality. At the end of this part, the effect of the motional information factor on video quality has been discussed. The influence sphere of an injury event which occurred in different frames of the same image group is determined by the motional information factor. According to this, we can calculate the effect of the motional information factor on frame image by calculating the ratio of the accumulated value of the time-domain diffusion factor and the number of total loss macroblocks.According to solve the second disadvantage of MPQM model, the TS parameters-based non-reference assessment model TSNM is proposed.This model not only take the number of the real network loss packets into account, but also calculate the number of the packets which can not be decoded caused by TS parameters error, that the video quality can be assessed more accurately. In the process of proposing TSNM, the TS packet format and the parameters in it are analyzed in detail firstly. Decoding of TS packets needs the program specific information (PSI). For the important effect of both the program association table (PAT) and the program map table (PMT) of PSI in the decoding of TS packets, the two tables are analyzed in detail. The above analysis provides a theoretical support for the proposing of the TSNM model. Then, the transmission rate of the TS stream is calculated by using MDI measurement method.The first and second priority TS parameters are analyzed in detail. Influence sphere, measurement and the assessment process of each parameter are given according to the characteristics of parameters. In particular, the first priority TS parameters play an important role in the process of decoding. So, the whole six TS parameters are analyzed, the number of TS packets which can not be decoded for TS parameter error is assessed by analyzing error type, duration and the number of errors. At the same time, the two of the second priority TS parameters are analyzed through the same method. According to analyzing quantitatively, the number of TS packets which can not be decoded due to parameter error is obtained. In the end, to improve MPQM model, TSNM model is proposed, and the assessment process is designed. The assessment result has five levels, correspond to different video quality of service.At Last, a prototype system is designed for testing MBNM and TSNM. There are two modules in the system: one is the information collection module, used for getting TS packets in detection point. The other one is quality assessment module, analyzes the TS parameters of the packet, used for assessing by TSNM model. Then, the TS packets are decoded to macroblock level, analyzed by the space-domain diffusion factor module and time-domain diffusion factor module. The number of total injury macroblocks can be get for assessing by MBNM model. A simulation experimental environment is built, and various network factors are simulated by injure software. According to experiment, these two models have good results in practical applications.