Image Stabilization via Local Motion Vector Accuracy Evaluation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image stabilization methods correct minority local motion vectors to align with majority vectors, often resulting in incorrect direction adjustments, leading to suboptimal image stabilization.
Innovation Solution
An image-processing method that evaluates local motion vectors using feature quantities, calculates accuracy of position adjustment, and corrects vectors based on evaluated values, ensuring accurate direction adjustments regardless of vector majority status.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Stability of the object's composition
If local motion vectors are corrected by aligning minority vectors with majority vectors, then the distribution of local motion vectors becomes smooth, but correctly-detected local motion vectors may be corrected in a wrong direction
Solution Approach 1:
The patent uses accuracy evaluation as feedback to determine whether to correct a local motion vector. Each local motion vector is evaluated for accuracy, and correction is performed only when the evaluation indicates the vector is erroneous. This feedback mechanism prevents correct vectors from being corrected in wrong directions while still smoothing the distribution by correcting actual errors.
Solution Approach 2:
The patent replaces the mechanical majority-vote correction system with an evaluation-based selection system. Instead of blindly following the majority direction, the system evaluates each vector's accuracy and selects the most accurate vector for correction, substituting mechanical consensus with intelligent evaluation.
2Ease of operation
If all local motion vectors are corrected based on majority alignment, then vector distribution becomes uniform, but the number of correct motion vectors decreases
Solution Approach 1:
The evaluation of each local motion vector provides feedback that determines whether correction is necessary. This feedback mechanism ensures that only vectors needing correction are modified, maintaining reliability while achieving uniform distribution through selective rather than universal correction.
Solution Approach 2:
The patent changes the parameter used for correction from simple majority alignment to accuracy evaluation. This parameter change allows the system to maintain vector correctness by using evaluation metrics to guide correction decisions, ensuring that uniformity is achieved without sacrificing reliability.
3Manufacturing precision
If local motion vectors are not corrected, then vector integrity is maintained, but image stabilization accuracy deteriorates due to erroneously-detected vectors
Solution Approach 1:
The patent segments the image into blocks and evaluates local motion vectors for each block individually. This segmentation allows targeted correction of erroneous vectors without affecting the entire image, improving stabilization accuracy while managing complexity through localized processing rather than global correction.
Solution Approach 2:
The patent uses neighboring local motion vectors as references to evaluate and correct erroneous vectors. By copying and comparing with adjacent vectors, the system can identify and correct errors without complex global analysis, improving accuracy while keeping the correction mechanism relatively simple.
Data Source
Figure 1~2
Figure 3~6
Figure 7~9
AI summary
An object of the invention is to provide an image-processing method, an image-processing program, an image-processing apparatus, and an imaging apparatus which can properly evaluate local motion vectors which are found by searching blocks of images upon detecting motion of plural images. To this end, the invention provides an image-processing method for detecting motion between plural images, including a searching operation of searching for a local motion vector from each of blocks in the plural images on the basis of feature quantity of the plural images, and a calculating operation of calculating one or plural evaluated values of the local motion vector found, in which one of the evaluated values is accuracy of position adjustment which is obtained when the search source block of said local motion vector is position-adjusted on a trial basis by using the local motion vector.