Underwater image restoration method based on dark channel prior and sparse representation

The invention discloses an underwater image restoration method based on dark channel prior and sparse representation. The method comprises the steps as follows: firstly, underwater image dark channel is calculated based on a dark channel prior theory, then the dark channel is processed with a denoising method based on a sparse representation model, solution of water transmittance can be optimized at the same time, and the transmittance is then refined by guided filtering; some pixels with higher brightness are selected from dark channel images, then gray value of the pixel with the highest brightness is found from a corresponding position of an original underwater image to be used as light intensity of water, and restoration of an underwater image is realized based on the solved water transmittance and light intensity of the water. The underwater image restored with the method is higher in definition, and noise contained in the underwater image is also better suppressed. 本发明公开了种基于暗原色先验和稀疏表示的水下图像复原方法，该方法首先利用暗原