Limited to the cost, the unmanned aerial vehicles are equipped with ordinary digital cameras for the main task, collecting the visible light image.In this paper, we proposed a method to identify oil and gas pipelines by color segmentation and shape detection.Firstly, we needed to set the ROI of the ROI, and calculated the covariance matrix C and mean m, and had image color cluster segmentation with Euclidean distance and Mahalanobis distance.Then, the edge detection was performed after that the image is segmented.Finally, the Hough transform is used to detect the line features according to the edge image to realize the automatic positioning of the pipe position in the complex environment.The test image bank contains 300 images, and the recognition accuracy is 80.3%.Experimental results show that the method based on color and shape features can effectively track and locate the pipeline in the background of large color difference.%限于成本,无人机搭载的任务设备主要为普通数码相机,采集的是可见光图像,针对此种情况,提出了一种利用彩色分割及形状检测识别油气管道的方法,首先需要设定感兴趣区域ROI,计算出协方差矩阵C和均值m,并使用欧氏距离、马氏距离对图像进行彩色聚类分割,然后对分割图像填色后进行边缘检测,最后根据边缘图像进行霍夫变换来检测直线特征,实现复杂环境下对管道位置的自动定位.测试图像库包含300幅图像,识别准确率达到80.3%,实验结果表明,在色彩差异较大背景中,基于颜色和形状特征的识别方法能有效进行管线跟踪定位.
展开▼
