Single-image reflection removing method and device and storage medium

Abstract

The invention discloses a single-image reflection removing method and device and a storage medium, and the method comprises the steps: obtaining a background image and a corresponding reflection imagethrough manual photographing, and obtaining a reflection image according to the superposition of the background image and the reflection image; inputting the reflective image into a pre-trained VGG-19 network for super-column feature extraction to obtain a feature set; inputting the feature set into a preset generation network to obtain a prediction background image and a prediction reflection image; inputting the prediction background image and the background image into a preset identification network to calculate an identification loss function of the identification network; carrying out iterative computation for many times until the joint loss function and the identification loss function of the generation network are converged, and completing the training of the generation network andthe identification network; and selecting a plurality of reflective images to carry out de-reflection processing so as to quantitatively evaluate the de-reflection effect. According to the method, high-level sensory information of the image can be extracted and added into training of the generative adversarial network, so that the de-reflection problem of a single image is effectively solved.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a single image dereflection method, device and storage medium. Background technique [0002] Reflection removal from a single image usually utilizes pre-specified prior information. First of all, the most common way is to use the sparse nature of the natural image gradient to find the minimized edges and corners to separate the layers of the image, such as using the constraints of gradient sparseness combined with the data fidelity term of the Laplace domain to suppress image reflections . However, this approach relies on low-level heuristics and is limited in cases where high-level analysis of image results is required. Another kind of prior knowledge is that the image of the reflective layer is usually unfocused and smooth. But algorithms based on this assumption cannot be applied to cases where the reflected image also has strong contrast. None of these me...

AI Technical Summary

Problems solved by technology

However, the algorithm based on this assumption cannot be applied to the case where the reflected image also has a strong contrast

These methods cannot effectively use the high-level sensory information of the image, and cannot solve the problem of dereflecting the reflective image with high contrast

Images