【摘要】 提出了一种新的红外图像分析算法。该算法根据图像预处理分割后的二值图像,对目标图像进行递归标记,避免了基于点的连通体检测中“1”像素点可能被重复扫描的现象,实现了只考虑提取目标特征,不用得到最终的标记图像本身的要求。获取对应红外图像的二进制编号后,根据目标形心特征进行变分辨率相关匹配,实验结果表明对理想及带噪声的输入均有较好识别率。该算法在提高了目标单帧检测概率同时,尽可能地降低了相关匹配的运算量。更多还原

【Abstract】 A new algorithm for the analysis of infrared image was proposed. Based on binary image made of image segmentation pretreatment, the recursion marking was assigned to the corresponding target. The algorithm avoids pixel "1" being rescanned in detection of the connected set of points, which realized the requirements of only considering abstracting the target characteristics without obtaining final marked image itself. According to the centroid feature of the target, it carries on correlation matching of alterable resolution by the corresponding binary number of the infrared image. The experimental results show that the algorithm has good recognition rate to the ideal input and the striping noise input. Moreover the algorithm improves the detection probability in the single frame. Meanwhile, the cost of computation of the correlation matching is decreased as far as possible.更多还原