Image Processing Motion Vector Detection Using Reliability-Based Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stabilization methods detect motion vectors with low reliability, requiring an increased number of blocks to improve accuracy, which increases operation complexity and may not precisely represent motion between images.
Innovation Solution
An image-processing method that searches for local motion vectors, elects representative vectors, and approximates motion vector distribution using a linear function, employing projected data and cumulative added images to reduce operation complexity and enhance precision without increasing the number of blocks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of blocks is increased to improve motion vector detection accuracy, then measurement precision improves, but device complexity and operation amount increase
Solution Approach 1:
The patent extracts only the reliable local motion vectors from the set of all local motion vectors by evaluating their reliability. Instead of using all blocks or increasing block number, the method selectively takes out the trustworthy motion vectors that meet reliability criteria, thereby maintaining detection accuracy without increasing operational complexity.
Solution Approach 2:
The patent introduces a reliability evaluation parameter to assess the quality of local motion vectors. By calculating reliability values and selecting vectors above a threshold, the method transforms the approach from quantity-based (increasing blocks) to quality-based (selecting reliable vectors), resolving the contradiction between accuracy and complexity.
2Measurement precision
If the number of blocks is increased to improve motion vector detection accuracy, then measurement precision improves, but the number of processing elements increases
Solution Approach 1:
The patent extracts only the reliable local motion vectors from the set of all local motion vectors by evaluating their reliability. Instead of using all blocks or increasing block number, the method selectively takes out the trustworthy motion vectors that meet reliability criteria, thereby maintaining detection accuracy without increasing operational complexity.
3Speed
If local motion vectors are used to represent motion between images, then detection speed improves, but reliability deteriorates due to low reliability vectors being mixed in
Solution Approach 1:
The patent segments the set of local motion vectors into reliable and unreliable categories by evaluating each vector's reliability. This segmentation allows the method to process only the reliable portion of vectors, maintaining detection speed while eliminating the harmful effect of low-reliability vectors on overall accuracy.
Solution Approach 2:
The patent introduces a reliability evaluation parameter to assess the quality of local motion vectors. By calculating reliability values and selecting vectors above a threshold, the method transforms the approach from quantity-based (increasing blocks) to quality-based (selecting reliable vectors), resolving the contradiction between accuracy and complexity.
Data Source
AI summary
An image-processing method is characterized by including a searching operation of searching for a local motion vector from each of blocks in the plural images, an electing operation of electing a representative motion vector from the local motion vectors of the respective blocks, and an approximating operation of approximating a motion vector distribution on the images based on the representative motion vector in order to detect motion between plural images with high precision without increasing the number of blocks made by dividing an image. In addition, when an approximate surface thereof is a plane, the operation amount for approximating the motion vector distribution and the information amount for representing the motion vector distribution can be minimized.