Realization And Optimization Of A H.264 Encoder On DM642

The H.264/AVC standard is a new generation video compression standard founded by ITU-T and ISO. There are four advantages with H.264. Firstly, it has a low bit-rate in comparison with MPEG2 compression technology. With the same quality, the data size generated by H.264/AVC is 1/8 less than that of MPEG2 code. Secondly, its highly sophisticated algorithms ensure rich details reserved even at the low bit-rates condition. Thirdly, it is highly fault tolerant, providing necessary tools to solve packet-loss and other errors caused by instability network Fourthly, it has network adaptive layer, making it more suitable for data transmission across different networks. The codec and the algorithms adopted in the implementation, however, are highly sophisticated.The main task of this paper is to introduce how to transfer and optimize the T264 encoder to DM642 platform. There are two levels to optimize the codec. The way of higher level use the conditional optimization provided by CCS and efficient function in IMGLIB, it is simple and effective; The way of lower level use intrinsics and linear assembly language, it improve the parallelism inherent in the H.264 encoder, but need deep understanding about DM642 and H.264.Our H.264 encoder can encode 35～50 frames per second for QCIF resolution video. And this encoder is used as a video server, it capture video from camera and send it to PC after code the image in DM642, At last a PC software will decode the data and display the video on screen.