In order to solve the difficulty of extracting feature points in the front-end of visual simultaneous localization and mapping (SLAM) for oil-immersed transformer robot, a standard deviation gray-world algorithm to enhance image under transformer oil is proposed in this paper. Firstly, in order to solve the problem of color distortion and low contrast of the image under transformer oil, this paper uses the traditional gray-world algorithm to enhance the image. Then in order to prevent the appearance of “excess” gain, this paper uses adaptive standard deviation to constrain the gray gain coefficient. Finally, the gray value adaptive stretching is used to solve the problems of color distortion and low contrast under the transformer oil. The experimental results show that the image under transformer oil processed by the algorithm proposed in this paper has bright color and more feature points extracted, the UIQM value of the enhanced image is increased by 65.09%, the UCIQE value is increased by 8.67%, and the number of feature points extracted is increased by 273.21%.