Method and device for automatically splicing image sequences and splicing system

Abstract

The invention discloses a method and a device for automatically splicing image sequences and a splicing system. The method for automatically splicing the image sequence mainly comprises: a step of detection used for detecting organization edges of adjacent images; a step of selection used for computing the similarity of the organization edge lines of the detected adjacent images, and selecting the edge line pair meeting a similarity threshold as an optimized edge line pair; and a step of acquiring splicing positions, namely determining the splicing displacement of the adjacent images in a horizontal direction and in a vertical direction according to the selected optimized edge line pair to determine the splicing positions in the horizontal direction and the vertical direction, wherein the splicing of the adjacent images is realized according to the acquired splicing positions. The method and the device are simply implemented and can effectively splice medical image sequences.

Description

technical field [0001] The present invention relates to an image processing technology, in particular to a method and device for automatically splicing image sequences to synthesize a panoramic image. Background technique [0002] Complete images of large-scale objects (such as X-ray images, or even complete X-ray images of human bodies) play an important role in orthopedic surgery, orthopedic or plastic treatment, etc. However, due to the limited size of imaging devices such as flat panel detectors (for DR systems) or imaging plates (for CR systems) in X-ray imaging systems, as well as other limitations of existing imaging techniques, only large-sized objects Take multiple body position shots, and then use image processing technology to synthesize the obtained body position images into a panoramic image. [0003] However, multiple shootings will cause the following problems: First, the exposure doses of two exposures in different body positions are different, resulting in ...

AI Technical Summary

Problems solved by technology

However, due to the limited size of imaging devices such as flat panel detectors (for DR systems) or imaging plates (for CR systems) in X-ray imaging systems, as well as other limitations of existing imaging techniques, only large-sized objects Take multiple body position shots, and then use image processing technology to synthesize the obtained body position images into a panoramic image

[0003] However, multiple shootings will cause the following problems: First, the exposure doses of two exposures in different body positions are different, resulting in inconsistent gray distributions in overlapping areas of adjacent images; Changes in voluntary movement, peristalsis of the large and small intestine, physiological contraction of the bladder, etc.) lead to differences in tissue images taken successively; third, changes in shooting conditions and other factors will cause slight distortion

These problems make it difficult to stitch images correctly

In addition, due to the large amount of data in medical images, the traditional grayscale or gradient global search is very time-consuming, which brings challenges to the dynamic and real-time application of medical images.

[0004] Taking X-ray images as an example, there are two main types of existing stitching technologies, one is to rely on special equipment to set obvious marks on each X-ray image, and then manually or automatically recognize the marks to achieve image stitching; the other The category is automatic stitching, one of which is dependent on the inherent characteristics of the device (such as the overlapping features of CR imaging plates), using different anatomical body images at multiple scales and using different algorithms or parameters to achieve stitching; the other is Image stitching technology that only relies on image features, but the accuracy is not high, often requires manual further correction and correction

In recent years, image panoramic stitching technology has developed rapidly. Although a technique using scale invariant feature transformation theory (Scale Invariant FeatureTransform: SIFT) for panoramic stitching has been proposed, it has not been successfully used in panoramic stitching of X-ray images. No report

Images