Splicing method of binocular stereoscopic image

Abstract

The invention discloses a splicing method of a binocular stereoscopic image. The method includes: S1, adopting binocular cameras to collect multiple groups of images, extracting features, and obtaining feature point sets by screening, wherein each group of images includes first eye images collected by a camera of a first direction, and second eye images collected by a camera of a second direction;S2, carrying out transformation and splicing on the first eye images according to the feature point sets, which are obtained by screening in the step S1, to obtain a first eye image after transformation and splicing; S3, calculating and obtaining a target parallax map according to the collected multiple groups of images; S4, carrying out transformation and splicing on the second eye images according to the first eye image after transformation and splicing and the target parallax map to obtain a second eye image after transformation and splicing; and S5, synthesizing the first eye image aftertransformation and splicing in the step S2 and the second eye image after transformation and splicing in the step S4 to obtain the final stereoscopic image. According to the splicing method of the binocular stereoscopic image provided by the invention, seamless splicing and ghost reduction can be realized, and limitation on placement positions and angles of the cameras is also eliminated.

Description

technical field [0001] The invention relates to the fields of computer vision technology and image processing, in particular to a binocular stereoscopic image splicing method. Background technique [0002] With the rise of VR and AR, people have higher and higher requirements for image resolution, viewing angle and quality. Due to the limited viewing angle range of a single camera, it is necessary to use image stitching technology to obtain wide viewing angle or even 360-degree panoramic images; currently Image stitching technology is one of the keys in VR and AR, and the application of image stitching is becoming more and more extensive, including medical care, education, sports, aerospace, etc. [0003] After taking multiple pictures with a single or multiple cameras, all the pictures are stitched into a wide-angle or panoramic picture. However, when a single photo is stitched into a wide-angle or panoramic picture, the depth of the scene from the camera to the picture is ...

AI Technical Summary

Problems solved by technology

[0003] After taking multiple pictures with a single or multiple cameras, all the pictures are stitched into a wide-angle or panoramic picture. However, when a single photo is stitched into a wide-angle or panoramic picture, the depth of the scene from the camera to the picture is unknown, making the stitching difficult. Blurring, ghosting, and misregistration issues in results

With the rise of stereoscopic images and videos, more and more people have begun to study stereoscopic image stitching technology. The existing stereoscopic image stitching usually has the following difficulties and problems: First, the processing of parallax. The existence of parallax makes splicing ghost images , affecting the final visual effect; the second is the splicing of randomly collected stereoscopic images. At present, some high-quality splicing image algorithms usually require the camera to be placed or rotated according to certain rules, for example, the camera needs to be fixed in a circle for shooting and splicing

Images