Keypoint Extraction for Robust Image Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image search technologies face challenges in accurately searching for similar images due to variations such as illumination changes, size changes, affine changes, and rotation, leading to decreased search speed and accuracy.
Innovation Solution
A server computing device and image search system that extracts keypoints from images, using preprocessing techniques like illumination compensation and normalization, and filters to robustly match keypoint descriptors, enabling efficient search for similar images across variations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If image-based search is performed on large amounts of image data, then search coverage is improved, but search speed decreases
Solution Approach 1:
The patent divides the image into multiple blocks and extracts keypoints from each block separately. This segmentation allows the system to process large images efficiently by working on smaller sub-regions, maintaining search speed while achieving comprehensive coverage across the entire image dataset.
Solution Approach 2:
The patent extracts key feature points and descriptors from images, separating the essential identifying characteristics from the rest of the image data. This extraction enables efficient comparison and matching operations, allowing fast search through large databases by comparing only the extracted keypoint features rather than entire images.
2Device complexity
If conventional image search is performed without preprocessing, then processing simplicity is maintained, but search accuracy deteriorates due to illumination and rotation changes
Solution Approach 1:
The patent applies preprocessing operations including illumination compensation and normalization before keypoint extraction. By performing these preparatory actions in advance, the system compensates for lighting variations and standardizes image characteristics, significantly improving matching accuracy without adding excessive complexity to the overall process.
3Speed
If keypoint matching is performed without verification, then processing speed is maintained, but matching reliability decreases
Solution Approach 1:
The patent implements a verification mechanism that checks the reliability of keypoint matches by analyzing geometric constraints and consistency across multiple keypoints. This feedback loop validates matching results and filters out false positives, ensuring high reliability while maintaining efficient processing through selective verification of candidate matches.
Data Source
AI summary
A server computing device and an image search system based on contents recognition using the same are provided. The image search system includes a server computing device, and the server computing device is configured to perform an operation of receiving image search basic information including at least one among an image, a keypoint extracted from the image, and an image search condition from a user computing device; an operation of extracting a keypoint which matches with a search request keypoint obtained from the image search basic information among keypoints which are previously stored; and an operation of transmitting at least one of an image and image related information corresponding to the extracted keypoint to the user computing device.