基于多光谱图像配准融合的机器人巡检平台

陈  麒1; 张  杰2

基于多光谱图像配准融合的机器人巡检平台

引用本文：陈麒1，张杰2.基于多光谱图像配准融合的机器人巡检平台[J].计算技术与自动化,2022,(2):118-124

摘要点击次数: 261

全文下载次数: 0

作者	单位
陈麒1，张杰2	(1.汕头职业技术学院机电工程系，广东汕头 515078 ２.汕头大学计算机系，广东汕头 515000)

中文摘要:针对变电站中开展机器人巡检工作时，普遍存在复杂场景下拍摄所得红外与可见光检测图像存在遮挡、旋转和视角差异等情况，进而导致图像匹配融合效率低的问题，本文提出了一种基于STM32控制的多光谱图像配准融合巡检机器人模型。该模型首先建立模板库，通过模板匹配定位图像中的目标电力设备，在此基础上使用SURF算法进行精细匹配，即利用RANSAC剔除设备特征误配点，并采用单映射FINDHOMOGRAPHY算法进行像素叠加，最终实现变电站设备的红外、可见光图像融合。实验结果表明，本文的改进算法生成特征点数量合适、质量提升明显，相比SIFT、SURF等传统图像融合算法，配准精度提高至少30%，图像处理速度提高至少30%，具备实用意义，可有效应用于设备识别、电力设备故障诊断领域。

中文关键词:设备图像配准融合模板匹配 SURF算法 RANSAC算法像素叠加

Inspection Robot Platform Based on Multi-Spectral Image Registration and Fusion

Abstract:When robot inspections are carried out in substations, it is common for infrared and visible light detection images captured in complex scenes to have occlusion, rotation and viewing angle differences, which leads to the low efficiency of image matching and fusion. This paper proposes a method based on STM32-controlled multi-spectral image registration fusion inspection robot model. The model first establishes a template library, and locates the target power equipment in the image through template matching. On this basis, the SURF algorithm is used for fine matching, that is, the equipment features are purified by RANSAC, and the single-mapping FINDHOMOGRAPHY algorithm is used for pixel overlay, and finally the substation equipment is realized Fusion of infrared and visible light images. The experimental results show that the improved algorithm in this paper generates a suitable number of feature points and a significant improvement in quality. Compared with traditional image fusion algorithms such as SIFT and SURF, the registration accuracy is increased by at least 30%, and the image processing speed is increased by at least 30%. It is of practical significance. It is effectively used in the field of equipment identification and power equipment fault diagnosis.

keywords:device image registration fusion template matching SURF algorithm RANSAC algorithm pixel superposition

查看全文 查看/发表评论 下载pdf阅读器