基于感兴趣区域的图像压缩编码技术已成为图像编码领域的研究热点。由于图像不确定性以及目标自动识别的实时性要求,现阶段通过模拟人类视觉特性确定感兴趣区域算法不能满足需要。基于目标红外图像的特点,提出了一种逐步逼近二维最大熵阈值递推搜索算法,综合运用二维最大熵阈值分割方法、顺序统计滤波、数学形态滤波等技术,实现了目标红外图像感兴趣区域的自动提取,经实验验证该方法不仅能满足一般的实时性要求,而且可靠、效率高,具有较高的实用价值。

Image compression based on Region Of Interest (ROI) is one of hot issues in the field of image compression and coding. Because of uncertainty of image content and real-time requirement of Automatic Target Recognition (ATR),the algorithm based on human visual system simulation,which is used in ROI automatic detection,cannot meet the requirements. Based on the characteristics of infrared target images,a new stepwise approaching and recurring threshold search algorithm based on two-dimensional maximum entropy principle was proposed. Through the algorithm above,order filtering and mathematical morphological filtering,the ROI automatic extraction about infrared target images was realized. An experimental study was also conducted. It is proved that the method of ROI automatic extraction can not only meet the real-time requirements,but also is reliable and effective,which has significant utility value.