一种有效的红外图像中人造目标分割方法

Effective segmentation of man-made target from IR image

提出一种红外图像单阈值分割方法。为了减少计算量,结合先验信息选择包含待分割目标的感兴趣区域,利用Bezier曲线法平滑感兴趣区域直方图的噪声;对平滑后的感兴趣区域的直方图求解其曲率曲线,用曲率曲线的波峰所对应的灰度值作为初始分割阈值;基于先验信息从初始分割阈值中确定最佳分割阈值并进行初步分割。为了弥补单纯利用阈值法分割的缺陷,结合上述分割结果和目标的边缘信息得到封闭性良好的完整目标的二值图像。实验结果表明,提出的方法能快速有效地将红外目标从复杂的背景中分割出来,算法的计算复杂度为O(MN)。

A new approach to segment man-made target from infrared image is proposed based on threshold segmenting method. A Region of Interest (ROI), which contains a target to be segmented, is extracted at first so as to reduce computation workload. And then smooth and denoise the histogram of ROI by Bezier curve method and segment it based on threshold method and prior information. The curvature curve of the smoothed ROI histogram will be obtained and the gray scale corresponding to peak of the curvature curve will be used as initial segmentation threshold. Based on prior information, the optimal segmentation threshold will be determined for initial segmentation. In order to remedy the faults caused by threshold segmentation, the above segmenting results and edge information of target should be integrated to obtain a binary image of target. Experimental results show that IR targets can be quickly and effectively segmented from complex background by the proposed method. The calculation complexity of this algorithm is O (MN).