Underwater images often suffer from severe color degradation, haze and local blur, which are attenuated by the scattering of suspended objects in water and the absorption of light by water. Aiming at the problem that the existing underwater image enhancement algorithms based on deep learning use a single convolution, up-sampling and down-sampling mode which leads to insufficient image feature extraction, this paper constructs the down-sampling module, up-sampling module and feature extraction module based on multi-scale feature extraction. On this basis, an underwater image enhancement network framework based on multi-scale feature extraction block cascade (MS-FEBC) is proposed. To further improve the feature extraction capability of the network, the CBAM attention mechanism is added to the high-dimensional feature space of the network. The experimental results demonstrate that compared with the existing algorithms, the algorithm in this paper effectively solves the problem that the underwater images have lower quality such as color-cast, hazing and detail loss. There is a significant improvement in all four objective evaluation indexes. The performance of the image SIFT feature point detection and Canny edge detection vision tasks is significantly improved.