一种红外点目标图像高保真压缩方法

针对红外点目标图像的特点,融合目标检测与无损-近无损压缩技术,提出了一种目标-背景分类的高保真压缩方法。将红外图像分成若干子块,对每个子块进行基于"最大中值滤波"背景抑制算法的点目标检测,根据检测结果将图像子块分为包含疑似目标的目标子块和不含目标的背景子块,对目标子块图像无损压缩,对背景图像子块近无损压缩,从而提高红外遥感图像压缩比,降低数据传输量,减轻数据传输压力。实验结果表明:所提的方法与传统全图无损压缩方法相比,能在不损失点目标信息的高保真压缩前提下,使图像压缩比提高40%以上。该方法对星上实时处理系统、红外探测跟踪系统的设计具有一定的理论和工程应用价值。

High-Fidelity Compression Method for Infrared Point Target Images

Combining the algorithm of point target detection and lossless or near-lossless image compression, as well as considering the character of the infrared point target images, a method of high-fidelity compression which compresses the target and background image patches separately is proposed in this paper. This method can reduce the transmitted data and relieve the contradiction between the limited bandwidth and the increased data without losing the information of infrared point target. This method partitions the infrared point target image into small patches, and detects the potential target point by using the maximum median filtering background suppression method. Then according to the detection result, the patch which contains the potential target is compressed in lossless mode, while the patch of background is compressed in near-lossless mode. Therefore it can achieve the high -fidelity compression. The experimental results show that compared to the traditional compression method, the proposed high-fidelity compression method can improve the compression ratio up to 40%. This method has important theoretical and practical value for the design of satellite-borne real-time processing system as well as infrared detection system.