Image super-resolution method and device based on light field acquisition device

Abstract

The invention discloses an image super-resolution method and device based on a light field acquisition device. The light field acquisition device includes a plurality of USB cameras and cameras to form a light field acquisition device with a 3×3 viewing angle. A plurality of USB cameras are regularly arranged in a square space and surround the cameras, wherein the method includes: acquiring multiple low-resolution images of the side view and high-resolution images of the middle view by collecting light field images; using dictionary learning and depth Learn to perform super-resolution on the collected low-resolution images of multiple side views; obtain the depth information of the scene according to the parallax between multiple light field images of different views. This method can predict and restore the high-frequency part of the input image through single-view and multi-view information, and use the information of multi-view high-resolution images to calculate the depth of the scene, which reduces the production cost and ensures the accuracy of spatial and angular resolution. .

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to an image super-resolution method and device based on a light field acquisition device. Background technique [0002] Light field acquisition and reconstruction technology is a very important issue in the field of computer vision. The use of light field for 3D reconstruction has great advantages over traditional 3D reconstruction methods: it is less dependent on hardware resources and is easy to carry out on a PC. Real-time reconstruction; strong applicability, the complexity of the scene has no effect on the complexity of the calculation. However, although high-precision 3D reconstruction can be performed by using a 3D scanner, its expensive equipment price and limitations in usage occasions limit its practical application. Light field technology has a wide range of applications in lighting engineering, light field rendering, re-illumination, re-focusing camera, syntheti...

AI Technical Summary

Problems solved by technology

In order to obtain a large angular resolution of the light field acquisition device, the parallax of images from different viewing angles is often relatively large. In this case, the dictionary learning method cannot effectively perform super-resolution, resulting in ghosting, blurring and other results, which greatly restricts the The accuracy of the information carried by the light field image, which further affects the accuracy of subsequent scene reconstruction, needs to be resolved

Images