基于灰度形态学和邻域熵值的弱小目标检测

Faint targets detection based on gray morphological filtering and neighborhood entropy method

提出了一种弱小目标检测的新方法。从实际应用出发,考虑到复杂的背景和大量的干扰噪声,对传统熵值检测算法进行了改进,采用邻域熵值变化为检测标准。为了提高此方法的有效性,结合了灰度形态学滤波来对图像进行预处理。该检测算法的全过程为:首先对图像进行形态运算;然后对形态波后的图像进行邻域熵的计算;接着以计算所得的邻域熵的最大值和最小值为依据对图像进行分割,得到目标或目标边缘所处位置;最后用实地拍摄的空中弱小目标真实图像进行了实验验证。结果发现:该新方法可对弱小目标、大目标、多目标进行检测,且检测速度快,抗噪声干扰能力强。

A new method of faint-target detection is put forward. To improve the practicality of the method, especially it is used to deal with those images that involves the complex background and a lot of noise, the method of neighborhood entropy was brought forward. Furthermore, to increase the validity of it, this method was combined with gray morphological filtering. First ,the image is processed with morphological filtering. Second, the neigborhood entropy is calculated. And then, the target is detected according to the maximum value and minimum value of entropy. The experiment results indicated the new method can detect small target stably and guickly.