In order to solve the problems of color distortion and low contrast in the images collected by the transformer detection robot in the deteriorated and discolored transformer oil, an image fusion enhancement algorithm for transformer oil is proposed. Firstly, the image is white-balanced by the perfect reflection algorithm to eliminate the influence of uneven illumination intensity under oil on the image color and make the color more balanced. Then, the color-corrected image is subjected to adaptive gamma correction to improve the contrast of the image. Finally, the multi-scale fusion strategy is adopted to fuse the image after color correction and the image after adaptive gamma correction, and a clear image under transformer oil is obtained. The experimental results show that the images processed by this algorithm are bright in color and rich in details. Compared with the original images, the image quality evaluation index (UCIQE), feature point matching and information entropy are significantly improved, which can provide clear data for transformer internal fault detection.