Image Change Detection via Sub-Block Segmentation and Background Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image analysis technologies face challenges in efficiently searching, classifying, and indexing objects within collections of images, particularly in detecting changes and associating them with backgrounds, which hinders effective object classification and event detection.
Innovation Solution
A method involving receiving images, estimating and validating backgrounds, detecting changes by subdividing images into sub-blocks, associating connected regions of changed pixels, and classifying these changes to index and store information for efficient retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If images are subdivided into sub-image blocks for change detection, then measurement precision of changes is improved, but device complexity increases
Solution Approach 1:
The patent divides images into sub-image blocks to perform change detection at a finer granularity level. This segmentation allows for more precise identification of changed pixels by comparing corresponding blocks between reference and current images, thereby improving measurement precision while managing processing complexity through localized analysis.
2Measurement precision
If connected regions of changed pixels are associated across multiple images, then object classification accuracy is improved, but loss of time increases
Solution Approach 1:
The patent performs preliminary change detection to identify connected regions of changed pixels before proceeding to full object classification. By pre-identifying regions of interest through change detection, the system reduces the search space for subsequent classification tasks, improving accuracy while mitigating time loss through staged processing.
3Measurement precision
If background estimation is performed for image collection, then detection of changes is improved, but device complexity increases
Solution Approach 1:
The patent performs background estimation as a preliminary step before change detection. By establishing a reference background model in advance, the system can efficiently detect changes by comparing current images against this pre-computed background, improving detection accuracy while managing complexity through separation of background modeling and change detection tasks.
Data Source
AI summary
Indexing regions of changed pixels (ROCHs) in a collection of images by receiving a collection of images. Estimate and/or validate a background among the collection of images. Detect the changes between images in the collection. Associate the detected changes between the images and classifying the associated changes. This image processing method and system may be used for image indexing and object classification.