基于图像拼接和帧间差分输电线路图像分割方法

Image Segmentation Method of the Power Lines Based on Image Stitching and Inter-frame Difference

针对无人机载红外摄像仪拍摄的输电线路红外图像易于受背景信息干扰且不易于分割出输电线信息,提出了一种基于图像拼接和帧间差分相结合的算法,来实现输电线分割定位信息。采用基于SIFT的图像配准算法来实现图像拼接,再将得到的两幅图像的重合部分进行帧间差分,初步实现了输电线的分割;采用Canny算法,对分割得到的输电线进行边缘检测,采用霍夫变换检测出分割出来的输电线,最终对输电线定位。

Infrared images of the power lines taken by infrared camera in UAV is susceptible to interference by background information and is not easy to split the characteristics from the power lines. The paper proposes an algorithm based on image stitching and combination of inter-frame difference, initially getting the power lines segmentation. Image registration algorithm which is based on the SIFT to achieve image stitching is used, and then the difference is got from the overlapping portion of the two images inter-frame, initially realizing the segmentation of the power lines. Then Canny algorithm is used to detect the edge of picture obtained by segmentation, the Hough transform being used to detect lines, and ultimately the power lines are located.