Aiming at mitigating the problems of low contrast, blurred details, and color distortion in underwater images, an underwater image enhancement method based on preprocessed image penalty and generative adversarial network (GAN) is proposed in this paper. First, an improved red channel histogram stretching algorithm is used to preprocess the input underwater image to improve the image contrast and avoid over enhancement of local blocks after traditional histogram stretching. Then, GAN with preprocessed image penalty is designed to realize underwater image enhancement. Moreover, multiscale convolution is used for the first two layers of the generator coding-decoding structure to enhance the detailed information learning ability of the network. Finally, a multiloss function is established in which the preprocessed image is used as a false truth value to impose loss penalty on GAN to improve generalization performance of the network. Experimental results show that compared with traditional image enhancement methods and deep learning-based image enhancement methods, the method performs better in terms of color deviation, contrast, and detailed information of underwater images, and has better robustness.