红外技术, 2019, 41 (12): 1111, 网络出版: 2020-01-09
一种复杂背景下的电力设备红外图像分割方法
Infrared Image Segmentation Method for Power Equipment in Complex Background
摘要
针对背景复杂的电力设备红外图像分割问题,提出一种新的分割方法。该方法运用线性谱聚类算法(LSC)对图像做超像素分割,将颜色、距离相似的像素聚类至同一个中心;利用在全局图像基础上计算所得的 Otsu阈值对各超像素块做背景预标记,并利用最大相似度区域合并算法(MSRM)对超像素块进行合并,在得到目标设备的同时,有效降低了过分割和欠分割率;最后运用数学形态学算法对图像做后处理,在保证设备特征的前提下提高目标设备分割精度。实验表明,在复杂背景下与其他算法相比,该方法可得到更为准确、完整的目标设备。
Abstract
A new segmentation method is proposed aimed at an accurate and complete solution for infrared image segmentation of power equipment with complex backgrounds. This method uses the linear spectral clustering algorithm(LSC) to obtain superpixel segmentation of images, which clusters pixels with similar colors and distances to the same center. Background premarking of each superpixel block is done under the Otsu threshold, which is calculated based on global images. The maximal similarity-based region merging algorithm(MSRM) is used to fuse the superpixels to obtain the target device and the over-segmentation and under-segmentation rates are effectively reduced. Finally, the mathematical morphology algorithm is used to post-process the image; this improves the segmentation accuracy under the condition of maintaining the characteristics of the device. Experiments show that this new method can obtain more accurate and better target equipment than other algorithms.
王小芳, 毛华敏. 一种复杂背景下的电力设备红外图像分割方法[J]. 红外技术, 2019, 41(12): 1111. WANG Xiaofang, MAO Huamin. Infrared Image Segmentation Method for Power Equipment in Complex Background[J]. Infrared Technology, 2019, 41(12): 1111.