To address the issue of insufficient identification of power equipment targets in aerial images, this paper utilizes grayscale conversion technology to convert color images into grayscale images. Noise reduction is achieved by applying Gaussian filtering algorithms and bilateral filtering algorithms to images of power equipment captured during drone patrols. Next, threshold segmentation and edge detection algorithms are applied to segment the images. The segmented binary images are then processed using adaptive morphological methods to eliminate interference introduced during binarization, thereby enhancing image readability and integrity. Results show that the proposed method achieves an accuracy rate of 96.91% for image segmentation precision and a correct rate of 97.5% for overall sample detection. The method achieves a fault detection accuracy rate of over 95% for power equipment images, demonstrating good performance in detecting corrosion-related faults. Additionally, even under complex background conditions, the model achieves an accuracy rate of over 90% in fault judgment. This highlights the significant theoretical and practical significance of this method in the intelligent detection application of power equipment.
