Abstract: According to the problem of infrared target which is difficult to identify in the complicated background and high requirement of real time, the infrared target segmentation operator is proposed, and the automatic image recognition mechanism is designed by introducing the characters matching. Based on H－S segmentation and image fusion, as well as morphology filtering and characters matching the problem of difficulty recognition is resolved. The solution with less calculation is implemented based on OpenCV with assembly optimization. And experimental results show that compared with other methods, this algorithm achieves good effect in automatic microbial image recognition, and is feasible in engineering.

Key words: infrared small target division operator, characters matching, H－S, OpenCV, assembly optimization, rapid recognition

摘要： 针对复杂背景下弱小目标识别困难，实时性要求高等难题，提出了弱小目标分割算子；并引入特征匹配，设计了二者相融合的自动图像识别机制。该方案基于H－S分割与图像融合，再结合形态学滤波与特征匹配，解决弱小目标难识别的问题。该方案计算量小并基于经过汇编优化的OpenCV实现，解决实时性要求高等难题。实验结果表明，相对于传统方法，该方案应用在自动图像识别上可以达到较好的效果，在工程上应用是可行的。

关键词: 弱小目标分割算子, 特征匹配, H－S, OpenCV, 汇编优化, 快速识别