Affected by the high turbidity, insufficient illumination and poor uniformity of the underwater environment, the images obtained by imaging mechanisms have defects such as low contrast, blurred details and color distortion. To handle above problems, an underwater image enhancement algorithm based on a dual-head enhancement and non-uniform correction was proposed. The dual-head enhanced network was constructed to extract multi-scale features from shallow information and to fuse the context information of different channels, which was conducive to the enhancement of low contrast of underwater images. Furthermore, the constructed non-uniform correction network was used to perform nonlinear weighted fusion of different channels and positions of the image, which was conducive to the recovery of color consistency and brightness. Compared with the 10 algorithms, the optimal value of the peak signal-to-noise ratio and structural similarity in the UIEB test set was improved by 4.02 dB and 0.120, decreased by 1.51 on the CIEDE2000 index, and decreased by 2.13 dB, 0.025 and 0.48 on the LUSI test set, respectively. Experimental results show that the proposed algorithm has a significant enhancement effect for non-uniform underwater images, and is more in line with the characteristics of the human eye.