Depth Map Estimation Method for Underwater Scenes Based on Underwater Light Attenuation Prior

Abstract

The present invention relates to an underwater scene depth map estimation method based on the underwater light attenuation prior, and the method includes the following steps: Step S1, underwater light attenuation prior; step S2, data set training; step S3, linear model Coefficient learning; step S4, scene depth map estimation. Its advantages are as follows: the present invention proposes a scene depth estimation method based on the underwater light attenuation prior, which mainly utilizes the underwater light attenuation prior, and selects a large number of training sets as sample training to obtain strong robustness. The underwater scene depth estimation model can quickly and effectively obtain the correct scene depth, and can be applied in the process of underwater image restoration.

Description

technical field [0001] The invention relates to the technical field of underwater images, in particular to an underwater scene depth map estimation method based on underwater light attenuation a priori. Background technique [0002] Underwater activities such as underwater archaeology, underwater resource exploration, and underwater pipeline exploration are facing great challenges, so obtaining clear underwater images plays a key role in ocean exploration. However, due to the influence of the underwater environment and its physical mechanism, underwater image processing has gradually attracted people's attention in recent years. In the imaging process of underwater images, the propagation of light in water is affected by the underwater environment, and there are problems such as selective absorption and dispersion, resulting in blurred underwater images, color casts, and low visibility. The attenuation of underwater images is mainly caused by factors such as different expon...

AI Technical Summary

Problems solved by technology

In 2010, Chao et al. directly used DCP to restore underwater images. This algorithm can only improve the contrast of some underwater images, but most of the restored images have color distortion.

In 2011, Yang et al. explored a simple and effective restoration method based on DCP. In order to reduce the computational complexity, the minimum filter and the median filter were used instead of soft matting (Soft Matting). This algorithm uses color correction to improve the quality of the restored image. Contrast, but low-quality restoration results limit the visual quality of the output image

[0004] The scene depth estimation method based on the physical model cannot meet the application of the actual environment, and requires a long processing time. This aspect needs to be optimized and improved.

At present, although the scene depth estimation methods based on DCP and UDCP can estimate the scene depth of underwater images of a small number of scenes, they cannot successfully estimate the scene depth for underwater images with complex scenes, because both DCP and UDCP use local block Search increases computational complexity and reduces operating efficiency

The underwater scene depth estimation method based on MIP considers the selective attenuation characteristics of underwater light in propagation and proposes a simple underwater scene depth, but this prior has not been fully proved and verified, so although this The method can optimize the calculation of scene depth estimation, but it cannot be applied to different types of underwater images

The underwater depth estimation method based on IMLA can obtain the scene depth of different types of underwater images, but the fusion method takes up a lot of calculations and cannot be applied in practical applications.

[0005] In summary, a priori of underwater light attenuation is needed to obtain a linear model for underwater scene depth estimation, which can easily and effectively estimate the scene depth map, and has a lower complexity underwater scene depth map estimation method, However, there is no report on this method

Images