Image Change Assessment via Keypoint Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image analysis technologies face inaccuracies in assessing object changes due to variations in image position, alignment, orientation, and background, as well as differences in materials, leading to unreliable comparisons between current and older images.
Innovation Solution
The method involves an image analysis computing device that retrieves training images and data related to the object, determines keypoints invariant to scale and rotation, and identifies changes by matching keypoints between the captured and training images, accounting for material differences and iterative learning to improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image comparison methods are used to assess object changes, then the process is simple and fast, but accuracy deteriorates due to variations in image position, alignment, orientation, background, and material differences
Solution Approach 1:
The patent segments the object into multiple parts and sub-parts, and further segments each part into triangular sections. This segmentation allows for localized analysis of changes in different regions, improving measurement precision by focusing on specific areas rather than treating the entire object as a single unit.
Solution Approach 2:
The patent transforms 2D image data into a 3D mesh representation by creating triangular sections from segmented parts. This dimensional transformation enables more accurate change assessment by adding depth and spatial relationship information, allowing the system to account for variations in position, orientation, and shape more effectively.
2Measurement precision
If detailed segmentation and analysis of parts and sub-parts are performed, then measurement precision improves, but processing time increases
Solution Approach 1:
The patent applies partial action by focusing analysis only on triangular sections where changes are detected, rather than processing the entire object uniformly. The system identifies changed regions and concentrates computational resources on those specific areas, reducing overall processing time while maintaining high precision in change detection.
Solution Approach 2:
The patent implements a dynamic processing approach where the level of segmentation and analysis adapts based on the detected changes. For regions with significant changes, detailed triangular segmentation is applied, while stable regions receive minimal processing. This dynamic adjustment optimizes the balance between precision and processing time.
Data Source
AI summary
A method, non-transitory computer readable medium, and an image analysis computing device that retrieves, based on a captured version of an object in a received image, training images which display related versions of the object and items of data related to the related versions of the object of the training images. Keypoints which are invariant to changes in scale and rotation in the captured version of the object in the received image and in the related versions of the object in the training images are determined. Changes to the object in the received image based on any of the determined keypoints in the related version of the object which do not match the determined keypoints in the captured version of the object are identified. The identified changes in the captured version of the object in the received image are provided.