Method and device of image processing before image stitching and method and device of image stitching

Abstract

The invention provides a method and a device of image processing before image stitching and a method and a device of image stitching. The method of image processing before image stitching includes: acquiring a to-be-processed image; determining a target area; invoking a two-dimensional linear scaling mechanism to carry out scaling processing of a first direction on the to-be-processed image, and using an image, which is output after processing of the two-dimensional linear scaling mechanism, as an image after scaling of the first direction; and invoking the two-dimensional linear scaling mechanism to carry out scaling processing of a second direction on the image after scaling of the first direction, and using an image, which is output after processing of the two-dimensional linear scaling mechanism, as an image used for image splicing. According to the method, scaling processing is carried out twice on the to-be-processed image through setting the two-dimensional linear scaling mechanism, thus correction for the image is realized, a code amount of correction processing is reduced, a programming algorithm and an execution process are simplified, image transmission time delay is reduced at the same time, and the system operating efficiency is improved.

Description

technical field [0001] The invention relates to the technical field of video image preprocessing, in particular to an image processing method and device before image splicing, and an image splicing method and device. Background technique [0002] At present, in terms of image display, users' demands for large screens, wide viewing angles, and low latency are gradually increasing. In order to meet the above requirements, multiple cameras such as 3 are often fixedly placed at a certain angle in image acquisition to increase the viewing angle range, and multiple such as 3 screens are used for splicing display to meet the above requirements of users. This requires the input images of multiple cameras to be seamlessly spliced ​​and displayed on multiple screens. The existing image stitching methods mainly include the following three categories: [0003] The first category: image fusion stitching method. This method uses the steps of feature point extraction, coordinate transfo...

AI Technical Summary

Problems solved by technology

Although this method is relatively simple to implement, there is still a problem that different correction processing mechanisms need to be used according to different images. For example, for images that are tilted in the horizontal or vertical direction, it is necessary to use a small-angle rotation mechanism to correct the image. Correction processing; for images captured by a camera with a pitch (left and right tilt) angle (the captured image is out of proportion up and down or left and right), an angle adjustment mechanism is required,

The implementation process of each of the above-mentioned mechanisms requires different processing mechanisms, which makes the amount of program code for image correction and splicing very large and the processing flow is many, resulting in slow image processing.

Images