| With the development of information and network technology, especially the mobile technologies, Global System for Mobile communication has been fully entered the era of3G. The3G era of technological progress, which makes the terminal access network bandwidth to further enhance and the mobile communication environment conditions have markedly improved. A variety of emerging, high-speed network-based value-added services and mobile multimedia services to mobile video on demand, mobile video telephony, video, SMS, mobile music, represented by the most promising, will become dominant in the future3G services. However, because the limited computing power of mobile terminals, as well as different operating systems and mobile phone standard, showing the diversity and heterogeneity of the characteristics; In addition, due to the instability of the mobile network, making the traditional video coding techniques can not be applied directly the mobile terminal. Therefore, the mobile video business in terms of how to effectively on mobile devices to achieve high compression efficiency and low computational complexity of video encoding method is of great significance. This paper in-depth analysis and research of key technologies for video encoding for mobile terminals, the main contents are as follows:1. Based on block characteristics of fast intra mode selection method research. Intra Prediction as an important means to improve the coding performance in H.264/AVC, it can effectively remove redundant information in the airspace of the original video sequence, but at the cost of complexity multiplied. In order to meet the needs of wireless networks and real-time applications, we use a method based on fast mode in the frame of the block features selected. Through the top and left of the current block, the respective four pixels were calculated variance to determine the texture changes to determine the appropriate mode, or deletion unrelated to the mode to enhance video encoding speed in the premise of ensuring the quality of video encoding.2. Hierarchical coding technology based on time-domain based on the AVS standard videoTime domain classification has very important significance for mobile video surveillance applications, mainly for the following reasons:First, the variety of mobile video surveillance terminal device, the device’s display resolution and the CPU processing capacity has significant differences, while video stream time-domain classification of sex can well fit a variety of terminals; Second, the mobile network bandwidth is not stable, when network bandwidth changes can not transfer the entire stream, it will give priority to reduce the video frame rate to adapt to the bandwidth, because the lower the resolution or image quality may lead to useful information in the image can not distinguish. Therefore, the AVS Working Group in the article "Technology needs for security applications", used for video surveillance, video encoding and decoding methods must support scalable domain classification. However, on the one hand, the B-frames will bring the encoding delay, and only two time-domain classification, on the other hand the B-frame encoding complexity is too high, not suited for mobile video surveillance applications. To solve these problems, this paper proposed a classification based on the AVS video domain coding techniques, including time-domain domain hierarchical domain classification in support of two non-reference frame P-frame technology and support multi-level classification coding technology, the former supports two frame rate, while the latter support the frame rate more than two time domain classification coding efficiency of AVS standard test, the test results show that:traditional IPP... P coding structure compared to the time domain classification coding efficiency of AVS standard decline is not obvious, even through the quantization parameter adjustment can improve the coding efficiency.3.Image quality evaluation model based on Contourlet transformImage quality assessment can be widely used for image acquisition, compression, restoration, enhancement processing system. Image quality assessment method based on the level of participation by the observer can be divided into two categories:subjective image quality assessment and image objective quality assessment. Subjective quality assessment accurately reflects the feelings of the human eye on image quality, demanding a rigorous testing environment, taking into account the large number of influencing factors difficult to control, steps to implement complex, costly, does not have real-time. Objective quality assessment method is the most commonly used method in image processing system, but the objective quality assessment algorithms in analog multi-channel decomposition of the eye image signal and CSF filtering, not yet take into account the representation of multi-scale features of the signal in several directions. To solve this problem, this paper presents the image quality evaluation model based on Contourlet transform. The algorithm Contourlet decomposition of the image signal to image multi-directional multi-scale characterization, sub-bands of different scales, the CSF threshold model to calculate the probability of each sub-band error can be perceived by the human eye and its sum, in order to establish a more accurate model of quality assessment. The experiments show that this algorithm with the objective image quality assessment algorithm based on wavelet, the assessment results with the results of subjective assessment better. |
