Image Stitching via Text Segment Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing algorithms face difficulties in accurately stitching two large images with overlap, making it challenging to obtain a reduced copy of the whole image.
Innovation Solution
An image processing apparatus comprising a text segment finding unit, a common text data finding unit, a feature point finding unit, and an image combining unit, which divides and detects text segments, extracts common text data, and combines images based on feature points to align and stitch images with overlap.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional image processing algorithms are used to stitch two large images with overlap, then the process becomes computationally complex and time-consuming, but the stitching accuracy remains poor
Solution Approach 1:
The patent segments the large images into multiple smaller blocks or patches. By dividing the images into manageable segments, the system can process each segment independently and efficiently, reducing the overall computational complexity and processing time while maintaining stitching accuracy through systematic matching of segment features
Solution Approach 2:
The patent performs preliminary actions by pre-processing images to extract key features, create feature descriptors, and establish initial matching relationships before the actual stitching process. This preliminary feature extraction and organization enables faster and more accurate stitching by avoiding computationally intensive operations during the final assembly stage
2Reliability
If conventional image processing algorithms are used to stitch two large images with overlap, then the algorithm complexity increases, but the stitching reliability remains low
Solution Approach 1:
The patent introduces intermediary elements such as feature descriptors, matching scores, and confidence metrics that mediate between raw image data and final stitching results. These intermediaries provide structured information that improves stitching reliability by enabling systematic evaluation and selection of correct matches while keeping the overall algorithm manageable through modular processing steps
3Ease of operation
If the whole large image is copied at once, then the copy operation is simple, but it is impossible to copy images that are too large to be copied at a time
Solution Approach 1:
The patent divides large images that cannot be copied at once into multiple smaller segments that can be processed individually. This segmentation approach maintains ease of operation by keeping each copy operation manageable while improving adaptability to handle images of any size through systematic division and subsequent reassembly of the segments
Data Source
AI summary
A method and apparatus for matching two images having areas of overlapping text, first image data, and second image data is provided. The method includes dividing the first image data into a plurality of scene segments and dividing the second image data into a plurality of scene segments, finding a text segment among the scene segments of the first image data and second image data, and detecting common text data in the text segments of the first image data and the second image data, the common text data having identical text data in the text segments of the first image data and the second image data. The method further includes extracting feature points from the first image data and the second image data based on the common text data, and combining the first image data and the second image data according to the feature points.