Image Processing Apparatus Partial Region Feature Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional imaging techniques face challenges in generating wide-angle panoramic images with high resolution, as the alignment of images with low feature amounts, especially from feature-less regions, often results in inaccurate motion vector detection and unnatural composite images.
Innovation Solution
An image processing apparatus with a determination unit, selection unit, motion vector detection unit, and composition unit is used to select images with sufficient feature amounts for alignment and composition, reducing the load of vector calculation by focusing on partial regions with higher contrasts or feature amounts, and performing template matching for accurate motion vector detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If motion vectors are detected from the entire images, then the alignment accuracy is improved, but the calculation load becomes excessively large
Solution Approach 1:
The patent divides the image into multiple partial regions and performs motion vector detection separately on each region. This segmentation approach reduces the overall calculation load while maintaining alignment accuracy by processing smaller, manageable sections rather than the entire image at once.
Solution Approach 2:
The patent applies different processing strategies to different regions based on their feature content. Regions with sufficient features are processed for motion vector detection, while regions with insufficient features are handled differently or excluded, optimizing the balance between accuracy and computational efficiency.
2Device complexity
If motion vectors are detected from partial regions only, then the calculation load is reduced, but the detection accuracy becomes insufficient when regions are feature-less
Solution Approach 1:
The patent evaluates the feature amount of each partial region and selectively applies motion vector detection only to regions with sufficient features. This local quality approach ensures high detection accuracy in suitable regions while avoiding the computational waste and potential failure in feature-less regions.
Solution Approach 2:
The patent changes the parameter of region selection based on feature amount thresholds. By dynamically determining which regions to process based on their feature content, the system adapts to varying image conditions and maintains detection accuracy across different scenarios.
3Area of stationary object
If all acquired images are combined regardless of feature amount, then the panoramic coverage is improved, but the composite image quality deteriorates due to inaccurate alignment
Solution Approach 1:
The patent applies quality-based selection to individual images based on their partial region feature amounts. Images with insufficient features in critical regions are excluded from combination, ensuring that only images meeting quality standards are used in the panoramic composite, thereby maintaining overall image quality.
Solution Approach 2:
The patent uses images with sufficient feature amounts as reference copies for alignment. By selecting high-quality images as references and matching other images against them, the system ensures accurate alignment and maintains composite image quality while achieving wide panoramic coverage.
Data Source
AI summary
An image processing apparatus determines whether a feature amount in a partial region of an image satisfies a preset condition, selects a second image to be combined with a first image from a plurality of images based on the result of the determination, and performs a composition process on the aligned first image and second image based on a motion vector between these images. From the plurality of images, the image processing apparatus selects as the second image at least any one of an image having a partial region with the feature amount determined to satisfy the preset condition and an image corresponding to a partial region out of a plurality of partial regions set in the first image with the feature amount determined to satisfy the preset condition.