Self-adaptive underwater image enhancement method based on differential evolution algorithm

Abstract

The invention discloses a self-adaptive underwater image enhancement method based on a differential evolution algorithm. The method comprises the following steps: performing histogram equalization operation on an underwater original image to obtain a color adjustment image; carrying out detail sharpening processing on the color adjustment image to obtain a detail enhancement image, carrying out brightness enhancement on the detail enhancement image based on a dark channel prior method, and carrying out defogging operation on a reverse image of the detail enhancement image to obtain a brightness-enhanced underwater image; an underwater image evaluation index is used as a fitness function of an optimization algorithm; selecting to-be-adjusted parameters, performing feedback adjustment on theto-be-adjusted parameters in the color adjustment, detail enhancement and brightness enhancement processes of the original underwater image through a fitness function of the optimization algorithm, and circularly executing the image processing process until a stop condition of the optimization algorithm is met, thereby obtaining an underwater enhanced image according to the optimal to-be-adjustedparameters.

Description

technical field [0001] The invention relates to the technical field of underwater image restoration, in particular to an adaptive underwater image enhancement method based on a differential evolution algorithm. Background technique [0002] With the continuous development of marine research, underwater image processing has become a relevant research in a wide range of fields, such as underwater telemetry, deep-sea exploration, etc. However, underwater images are susceptible to blurring due to absorption and scattering in water, resulting in reduced contrast and grayed out colors. This can compromise computer vision applications, especially security-critical ones. Researchers have investigated ways to transform images into better representations or improve their visual appearance to benefit subsequent image processing. Some existing techniques utilize multiple images or depth information to improve image quality. There are other techniques that use histogram-equalization-b...

AI Technical Summary

Problems solved by technology

Although deep neural networks perform well on image enhancement problems, they are very sensitive to the settings of their hyperparameters

Finally, the performance of boosting algorithms is often affected by many hyperparameters, which are unknown and difficult to choose in practice

Although there are a large number of advanced augmentation methods studied using rules of thumb, it is difficult to find excellent empirical agreement between good augmentation performance and hyperparameter settings

Images