Research on the Infrared and Visible Power-Equipment Image Fusion for Inspection Robots

In this paper we present a novel and robust fusion algorithm for the infrared and visible Power-Equipment image, which is invariant to large-scale changes and illumination changes in the real operating environment of Power Equipments. Firstly, the Scale-Invariant Feature Transform (SIFT) algorithm is used to extract and describe the feature points from the infrared and visible images. Secondly, the best feature matching for each feature point of visible images is found by identifying its nearest neighbor in the database of feature points from infrared images. Thirdly, the Random Sample Consensus (RANSAC) technology is chose to select the appropriate geometric transformation model and estimate the transformation parameters of this geometric model. Finally, the bicubic interpolation method is employed to implement grayscale interpolation and coordinate transformation, then obtain the fusion image of visible and infrared images. Extensive experimental results on the Power-Equipment dataset demonstrate that our method has high stability and excellent performance.