Research Of Android-Based Tablet Video Quality Evaluation System

With the rapid development of mobile internet, the three mainstream systems iOS, Android and Windows have already penetrated the mobile terminal device. Android system has attracted a large number of developers by its open-source. In the development of various forms of mobile terminals, App Store is colorful, including mobile video that has attracted people of all ages. From analog to digital video, video sampling, compression, transmission and reconstruction process will produce different degrees of distortion. At present, the majority of video quality assessment is based on the PC, but relatively little research in the field based on mobile terminals. The focus of future research is that how to improve the efficiency of terminals-based video quality assessment.This program has been studied mainly two aspects. On the one hand is the design of Android-based yuv video player. It needs to call the third-party libraries SDL to realize the play of yuv formats. On the other hand is the development of Android-based video quality evaluation system, which is mainly composed of subjective evaluation and objective evaluation. Subjective evaluation used ACR-HRR method defined in standard ITU-T P.910. Objective evaluation included peak signal to noise ratio, point sharpness method, the edge structural similarity and Spatial Information-Time Information as four index algorithms. Using five video scenes provided by VQEG as testing sequences, every sequence consists of one reference video sequence and sixteen distortion video sequences. The system was implemented through Java programming language. First, calculating the index values by system. Then using entropy weight method to calculate the weight. Third, constructing the utility function to calculate the Euclidean distance between each sequence. Finally, making the quality order according to the Euclidean distance to finish the video quality evaluation.The results show that in the five scenes the quality of distortion sequence 16 is the best and the distortion sequence 9,10,15 are higher than others. In the meanwhile, the worst in every scene is distortion sequence 1, the quality of distortion sequence 1,3,11 in every scene is worse. In the whole, the quality of sixteen distortion video sequences have the same trend. Although the objective evaluation and subjective evaluation have some differences, however, the comprehensive evaluation agrees with subjective results with good consistency. It proves the advantages of multi-index evaluation. The results of this test objectively reflects the quality of the video sequence. It will be a significant guiding for video quality assessment based on Android tablets and other mobile devices.