改进的红外弱小目标检测方法

摘要

为了进一步提高红外弱小目标检测的准确性和实时性,针对红外图像的特点利用修正的Top-Hat算子对目标图像进行背景抑制以提高目标检测概率.根据已有的先验知识构造属性集,把灰度直方图限定在感兴趣区域,减少背景和噪声的干扰.利用改进的OTSU算法进行图像分割以检测目标.为了提高分割算法运算速度,把快速递推算法推广应用到了本文算法中.实验选取了不同类型的红外图像,将本文算法和其他文献中算法的分割结果进行了比较.实验结果表明,本文算法具有更好的分割效果,目标分割准确,运算速度提高了60％以上.属性直方图可以针对具体问题,利用已有的先验知识灵活构造,具有较好的通用性.%In order to further improve the accuracy and real-time performance of small infrared target detection, a new improved method is proposed. First, aiming at the characters of the infrared image, the improved Top-Hat filter is used to restrain background for the purpose of increasing the target detection probability. Second, the bound set is constructed to limit the gray level histogram to region of interest for reducing the interference of background and noise. Then target detection is achieved by image segmentation using improved OTSU algorithm. Furthermore, the fast recurring algorithm is applied to proposed algorithm for accelerating the running speed of the segmentation algorithm. Different types of infrared images are selected to compare the results of purposed algorithm and the other two algorithms in references. Experiment results show that proposed algorithm has better performance than the other two markedly. Object segmentation is more accurate and running speed is accelerated by more than 60%. Finally, the bound histogram can be constructed aiming at concrete problem, based on gained prior knowledge, so it has better generality.