Adaptive enhancement of underwater image based on dense residual denoising

Due to the particularity of underwater complex environment, the small targets in underwater images cover impulse noise, speckle noise and other causes, resulting in low image resolution, single detail expression, low information entropy and average gradient. Therefore, this paper proposes an adaptive enhancement method for underwater small target images based on dense residual denoising. Combined with dense residual network-denoising technology, the feature extraction structure of underwater small target image is constructed, and the impulse noise part of the image is removed by median filtering, and then the speckle noise part of the image is removed by wavelet filtering to denoise the underwater small target image. Through the method of multi-scale averaging, the features of the enhanced image are in a balanced state, and the Sigmiod function is selected as the objective function of the image. By integrating dense residual network and adaptive enhancement strategy, the underwater small target image is enhanced. The experimental results show that the resolution of the image is significantly improved, the clarity is higher, the details are rich, and the information entropy and average gradient are higher after the image is enhanced by this method, thus achieving a better visual presentation effect of underwater small target images.