Research on a wavelet transform based hybrid rate allocation algorithm for stereo images compression

Three regional images could be obtained by the three-line-array stereo CCD camera with three sensors named forward-sensor, nadir-sensor and back-sensor at the same time. The amount of data got by this kind of camera is three times to the single camera system, so it´s difficult to transmit the data in real time over a satellite to ground channel. At present, the stereo images are compressed independently, without using the characteristics of the stereo image features, so that the compression efficiency is lower. In this paper, a novel stereo image compression algorithm using a wavelet transform based hybrid rate allocation is proposed. The three stereo images could be compressed with different compression rate which was adaptively allocated according to each image complexity with the restriction on the total compression rate,and the reconstructed images would be less distortion. The experiment results show that the algorithm is better than the single band compression either on the PSNR, geometry precision or the subjective quality. The quality of the reconstructed images from the forward-sensor, nadir-sensor and back-sensor is more balanced, which is conducive to the subsequent processing and applications of the mapping data.