Design And Implementation Of Compression Algorithm For Aerial Images In Low Bit-Rate Based On Davinci

Aerial images have been used more and more widely. And higher efficient image compression algorithm is required. However, the existing compression algorithms do not meet the demand. According to the characteristics of Davinci platform, this thesis designs a compression algorithm for aerial images compression in low bit rate based on JPEG2000 framework. The main research work of this thesis is as fellows:(1) A wavelet transform algorithm based on windows is proposed for Davinci hardware platform. The algorithm can reduce the memory requirement greatly. The results of experiment show that this algorithm decreases the data exchange between in and off chip memory and speeds up image coding.(2) A new method for wavelet coefficient processing is proposed. Through analyzing the statistical law between quantization steps and compression ratios, a quantization step design algorithm is proposed to reduce time overhead. And a bit-plan shift method is proposed to preserve the edge information.(3) A rate control algorithm based on bit-plan priority is proposed. The results of experiment show that the algorithm can decrease redundancy encoding and time consumption.(4) The algorithm is optimizationed for Davinci platform. DMA and parallel pipeline is used to optimize the implementation of image coding. In order to make the code more suitable for Davinci platform, a part of the code is rewrited in assembly.As a conclusion, the aerial images can be compressed real-time, and the edge information can be preserved better. The proposed algorithm can meet the demand for image compression.