Substation remote infrared image registration based on multi-scale Harris corner and hierarchical guiding match strategy

The sequence remote infrared images acquired from the intelligent substation monitoring system existed position deviation. The deviated images need to be registered in order to ensure the system accurately locating and monitoring the interested equipment. This paper has proposed a substation remote infrared image registration method based on multi-scale Harris corner and hierarchical guiding match strategy. Firstly, multi-scale Harris corners are detected in each layer of the image pyramid, and using an interpolation location approach to improve the accuracy of the corner position. Further, the corners are refined to make them well-distributed in the image space. In the matching stage, adopting different registration method in different layers to fulfill a coarse to fine registration. And the initial registration result in upper layer is transferred to the lower layer to guide the new corner matching. This matching approach greatly improves the efficiency. The final experiment indicates that our method has acquired high registration accuracy and short time and good stability, all controlled within 6~7 seconds. The algorithm can be applied to the actual system.