Aiming to address issues such as blue-green color cast, haze, and noise in underwater images, a color-line model-based and multi-scale fusion-based underwater image enhancement algorithm is proposed. At the first stage, a local adaptive color correction method is designed to preprocess the degraded image to improve the color line fitting effect and reduce the degree of image color bias. Afterwards, a convex optimization algorithm is constructed to estimate the transmittance by solving the bias using the color line and the background light vector, and the accurate model parameters are obtained to recover the image to achieve the image defogging. Additionally, the white balance algorithm is used to compensate for the depth-selective absorption of color deviation in the pre-processed image. Meanwhile, the white balance processed image is processed by gamma correction and noise suppression algorithm based on color line constraints to improve the global contrast and suppress the noise, respectively. Ultimately, multi-scale fusion of the defogged, contrast-enhanced and noise-suppressed images is performed to obtain a feature-rich underwater enhanced image. Experimental results show that the proposed method can effectively solve the phenomena of color deviation, atomization and noise in underwater images. Compared with the comparison algorithms, the underwater color image quality evaluation index and peak signal-to-noise ratio are improved by 18.37% and 42.16%, respectively, and the enhancement results can better retain the underwater color image and reduce the image noise.