Image Processing Apparatus Motion Vector Detection Grid Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stabilization methods for video cameras face challenges in real-time motion vector detection due to uneven feature point distribution and data handling issues, leading to delayed processing and increased memory data requirements.
Innovation Solution
An image processing apparatus and method that divides frame images into grid areas, determines representative points for each area, and sets image portions for motion detection based on these points, allowing for efficient correlation-based motion vector detection without relying on external devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If feature point extraction is performed on an entire image, then the motion vector detection rate is improved, but the distribution of feature points becomes uneven causing image stabilization to be focused on areas with concentrated features
Solution Approach 1:
The image is divided into multiple grid blocks, and feature point extraction is performed independently in each grid block. This segmentation ensures uniform distribution of feature points across the entire image while maintaining sufficient feature points in each local region for accurate motion vector detection.
2Measurement precision
If the template block and search range are set to protrude from grid block boundaries, then the motion vector detection accuracy is improved, but the processing time is increased due to delayed template matching
Solution Approach 1:
Image data in extended regions that may be needed for template matching is pre-loaded into memory before template matching is performed. This preliminary action eliminates waiting time during template matching operations, allowing the template block and search range to protrude from grid block boundaries without delaying processing.
3Measurement precision
If the template block and search range protrude from grid block boundaries, then the motion vector detection accuracy is improved, but the data readout amount from memory is increased
Solution Approach 1:
The memory structure is organized with separate storage regions for different grid blocks, allowing the system to selectively read out data only from the specific extended regions needed for each template matching operation. This local quality approach reduces unnecessary data readout while maintaining the ability to access protruding template and search ranges.
Data Source
AI summary
An image processing apparatus comprises: a dividing unit that divides two frame images into a plurality of divided areas; a determination unit that determines a representative point for each of the divided areas in one of the two frame images; a setting unit that, for each of the two frame images, sets image portions for detecting movement between the two frame images, based on the representative points; and a detection unit that detects movement between the two frame images based on correlation values of image signals in the set image portions, wherein for each of the divided areas, the determination unit determines a feature point of the divided area or a predetermined fixed point as the representative point of the divided area in accordance with a position of the divided area in the one frame image.