Improved Sparrow Search Algorithm and Its Application in Infrared Image Segmentation

Faults in power equipment are often observed during inspections as abnormal heat through infrared image detection. To address the problem of poor accuracy and efficiency in thermal fault diagnosis of power equipment using the Otsu method, an infrared image segmentation method based on variable spiral sparrow search algorithm (VSSSA) is proposed. VSSSA first uses tent chaotic sequences to improve the initialization. Then, Lévy flight and variable spiral strategy were introduced to enhance the optimization speed and exploration ability of the population. The effectiveness of the algorithm performance was verified using benchmark function tests. Finally, on the basis of VSSSA optimization of the two-dimensional Otsu function and double threshold segmentation of infrared images combined with adaptive region growth method, the accurate target region was further extracted. The experimental results of image segmentation demonstrated better accuracy of the proposed algorithm compared with that of other segmentation methods. This has certain practical applications.