基于运动补偿和码率预分配的干涉多光谱图像压缩算法

Compression of Interferential Multi-Spectrical Image Based on Motion Compensation and Rate Pre-Allocation Algorithm

提出了一种基于运动补偿的三维小波变换和基于码块预测的码率预分配的图像压缩算法.利用干涉多光谱图像成像推扫平移特性,在小波变换中使用运动补偿来减少帧间相关性,并对图像组中各个图像小波变换和量化后EBCOT编码码块的有效比特平面进行独立的熵估计.以图像估计熵总和指导整个图像组码率预分配,以解决平均分配码率对重建图像质量带来的影响.实验结果表明:该算法在8倍压缩时,图像序列的平均峰值信噪比比3D-SPIHT提高了0.85～1.25dB,比单帧JPEG2000提高了1.91～4.25dB,算法复杂度低,易于硬件实现.

A new multi-spectral image compression algorithm combining three-dimensional wavelet transform based on motion compensation and rate pre-allocation algorithm based on code block prediction was proposed. In order to reduce strong correlation between image series, motion compensation was used in this wavelet transform. After wavelet transform and quantification,a prediction module was introduced to estimate the entropy of EBCOT code block within available bit plane,then the summation estimate entropy of each image was used to allocate the rate for each image to eliminate influence of the reconstructed image quality using equal allocation. The experimental results show that the proposed algorithm outperforms 3D- SPIHT 0. 85～1. 25 dB in average PSNR of image series with eightfold compression. Compared to JPEG2000 algorithm,the proposed algorithm improves 1.91～4.25 dB. The proposed algorithm is suitable for hardware implementation with low computation.