Image splicing method and system

Abstract

The invention provides an image splicing method, which comprises the following steps of: firstly, performing fixed-point rotation or moving shooting along a direction vertical to an optical axis of a camera to obtain a first source image and a second source image; carrying out vector change on the first source image and the second source image to obtain a corresponding vector matrix, and extracting feature points from the image matrix by adopting an improved SIFT algorithm to obtain corresponding feature point coordinates; selecting a main direction of a feature point, and generating a feature vector; rough matching is carried out on feature points in the feature vectors based on constraint conditions to obtain feature points after rough matching, and then the feature points after rough matching are matched by adopting an RANSAC (Random Sample Consensus) algorithm; splicing the images by adopting an overlapped part linear smooth fusion method to obtain a panoramic image; according to the method provided by the invention, the splicing speed can be increased while the splicing effect is ensured, smooth transition can be realized in the overlapping region in the final image fusion step, and the seamless splicing effect is achieved.

Description

technical field [0001] The invention belongs to the field of image processing, in particular to an image splicing method and system. Background technique [0002] Image stitching is to fuse multiple images with a certain overlap to synthesize a high-resolution image with a wide field of view and no obvious stitching gaps. At present, image stitching technology is widely used in computer image processing, seabed exploration, medical image, virtual reality, remote sensing image processing and other fields. [0003] Image mosaic algorithms can basically be divided into the following two types: methods based on grayscale information and methods based on features. The method based on grayscale information calculates the similarity between two images by using the pixel values ​​of the images to be stitched, so as to determine Stitching overlapping areas to achieve image stitching, this method is computationally intensive and less robust. The feature-based method obtains the mapp...

AI Technical Summary

Problems solved by technology

[0003] Image mosaic algorithms can basically be divided into the following two types: methods based on grayscale information and methods based on features. The method based on grayscale information calculates the similarity between two images by using the pixel values ​​of the images to be stitched, so as to determine Stitching overlapping areas to achieve image stitching, this method is computationally intensive and less robust

The feature-based method obtains the mapping relationship between images by extracting useful feature information on the image, and then matching the features of the two images. This method has high robustness, and most stitching algorithms currently use This method, but the algorithm is too dependent on the gradient direction of local pixels, and the layer-by-layer pyramid difference construction of the image is easy to cause the problem of feature point positioning deviation, and it takes a long time to complete the image stitching task efficiently.

Images