Image Stabilization via Feature Point Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Camera vibration interference causes displacement of target objects in images, affecting recognition and positioning in image recognition systems.
Innovation Solution
An image processing method and apparatus that extracts feature points from key and to-be-processed frames, determines matching relationships, and applies a transformation relationship to align the angle of view, thereby stabilizing the target object's position in the image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If camera vibration interference is present during image capture, then image acquisition is possible, but target object displacement occurs affecting recognition and positioning accuracy
Solution Approach 1:
The patent extracts feature points from a key frame image before processing subsequent frames. By establishing feature point correspondence between the key frame and to-be-processed frames in advance, the system can compensate for vibration-induced displacement through transformation relationships, thereby maintaining positioning accuracy despite camera vibration during capture
Solution Approach 2:
The patent creates a transformation relationship that maps feature points from the to-be-processed frame to the key frame coordinate system. This virtual copying of the key frame's stable coordinate system to the vibrating frame allows accurate target positioning without physically stabilizing the camera during capture
2Measurement precision
If feature point extraction and matching is performed to correct vibration displacement, then target object positioning accuracy improves, but image processing complexity increases
Solution Approach 1:
The patent extracts only the essential feature points from images rather than processing all pixel data. By identifying and matching only salient feature points between frames, the system achieves vibration compensation with reduced computational complexity compared to full-image processing methods
Solution Approach 2:
The patent applies different processing strategies to different parts of the image: feature point extraction and matching is performed only at key locations rather than uniformly across the entire image. This localized approach maintains positioning accuracy while reducing overall processing complexity
Data Source
AI summary
Embodiments of the present disclosure disclose an image processing method and apparatus, and an electronic device. The method includes: obtaining a key frame image that includes a target object, and obtaining a to-be-processed frame image that includes the target object; extracting a feature point in the key frame image and a feature point in the to-be-processed frame image, respectively; determining a matching relationship between each feature point in the key frame image and each feature point in the to-be-processed frame image; determining a transformation relationship based on the feature point in the key frame image and a feature point in the to-be-processed frame image that matches the feature point in the key frame image; and processing the to-be-processed frame image based on the transformation relationship to obtain a target frame image.