Feature Point Positioning Using Reference Image Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional feature point positioning technologies face inaccuracies when the target area is blocked, leading to suboptimal results in applications like face recognition or expression recognition due to the inability to accurately locate feature points amidst obstructions.
Innovation Solution
A method and apparatus for feature point positioning that utilize a reference image to determine image feature differences and target feature point location differences, enabling more accurate positioning by using a feature point location difference determining model and image feature extraction models to refine the positioning of target feature points in current images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional feature point positioning technology is used, then the positioning process is simple, but the positioning accuracy deteriorates when the target area is blocked
Solution Approach 1:
The patent extracts feature points from a reference image (previous frame) in advance and stores them. When positioning the current frame, these pre-extracted feature points are used as a basis for comparison and tracking, avoiding the need to re-extract all feature points from scratch and improving accuracy even when the target is partially blocked.
Solution Approach 2:
The patent establishes a feedback mechanism where feature points from the reference image are compared with feature points in the current image. The system calculates feature point location differences and uses this feedback to refine positioning, continuously improving accuracy across multiple frames even under occlusion conditions.
2Reliability
If feature point positioning is performed without reference to previous frames, then the processing speed is fast, but the positioning stability deteriorates due to jitter
Solution Approach 1:
The system performs feature point extraction on reference images in advance and stores the results. When processing the current frame, it only needs to compare with the pre-extracted feature points, significantly reducing processing time while maintaining stable tracking across frames.
Solution Approach 2:
The patent maintains continuous tracking by constantly comparing feature points across multiple consecutive frames. This continuous comparison smooths out random variations and jitter, providing stable tracking over time while the efficient comparison process minimizes time loss.
3Measurement precision
If only the current image is used for positioning, then the processing is simple and fast, but the positioning accuracy deteriorates when the target area is blocked
Solution Approach 1:
The patent creates a copy of feature points from the reference image (previous frame) and uses this copy as a template for comparison with the current image. This copying approach allows the system to leverage information from unblocked areas in the reference image to accurately locate feature points even when they are blocked in the current image.
Solution Approach 2:
The patent transitions from single-frame positioning to multi-frame positioning by incorporating temporal dimension. It compares feature points across different time frames (reference image and current image), adding a temporal dimension to the positioning process that provides redundancy and improves accuracy when the target is blocked in any single frame.
Data Source
AI summary
This application relates to feature point positioning technologies. The technologies involve positioning a target area in a current image; determining an image feature difference between a target area in a reference image and the target area in the current image, the reference image being a frame of image that is processed before the current image and that includes the target area; determining a target figure point location of the target area in the reference image; determining a target feature point location difference between the target area in the reference image and the target area in the current image according to a feature point location difference determining model and the image feature difference; and positioning a target feature point in the target area in the current image according to the target feature point location of the target area in the reference image and the target feature point location difference.