| Data processing is the most common way for the monitor, but this method can not give people the intuitional scene for the running of the manufacturing process. The original data processing and video surveillance can give people both the data message and video message. And microcosmic message can get from image processing.The data of the original image is very large and not very suitably for transmission, it is need to find a method to compress the video data, especially for the embed system which have the limit resource. The ITU-T and ISO/IEC have released the new video compress standard, and this new standard is named H.264 or the 10th section of MPEG-4. The notion of the new standard is more or less the same as the former, but it can supply a meeting the requirements, reliable and efficient video in different network. This paper combine the TI processor and put forward a solution which achieve the collection, encoding, transmitting system of video by using the DSP. The project comes from the collaboration of lab and Jiuduan company of Beijing.After analyzing the status and superiority at home and abroad of the H.264, this paper analyses the levels and performance, and decides to adopt the x264 to start our project. Then this paper analyses the encoding tools of the H.264, also analyses the cost of these tools to see witch tool should be optimized most. After the former analyzing, it is obviously that motion estimation cost most. This paper combines the MVfast and ME-skip, and designs a motion estimation method based on the ME-skip, and from the test result can be seen that the capability of the encoding is enhanced. Last, the paper optimizes the code based on the chipset. Also, this paper uses the DSP/BIOS to design a capturing, encoding, transmitting system of video, the SCOM is used in the communication between the tasks. Taking the advantage of the TI's NDK, designs a transmission of the video base on the UDP. Optimizing the c code is the needed to enhance the capability of the encoder. Firstly reasonably distribute the cache and memory, choose the reasonable option for the compiler to achieve the best effect. Writing the liner assembler for the most cost section of the code is the last step for the project. From the test, the capability of this project is: 44 f/s QCIF for real-time capturing, encoding, transmitting for video. |
