Stereo Image Analysis System for 3D Shape Restoration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for restoring a three-dimensional shape of a photographing object from stereo images are inefficient due to inaccurate extraction of corresponding points and insufficient detail in depth information, particularly affecting glaucoma diagnosis in the medical field.
Innovation Solution
An image analysis system that captures and analyzes stereo images by assigning weighing factors based on pixel contrast size, calculating local contrast, and determining similarity between local areas to accurately extract corresponding points and compute depth information, using a computer with a CPU and memory for precise three-dimensional shape restoration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional methods are used to extract corresponding points from stereo images, then the process is simple, but the extraction accuracy is insufficient leading to poor three-dimensional shape restoration
Solution Approach 1:
The patent divides the image into multiple local areas around corresponding points and analyzes each local area separately. This segmentation allows for detailed contrast analysis in specific regions, improving corresponding point extraction accuracy without requiring the entire complex image to be processed uniformly.
Solution Approach 2:
The patent calculates contrast values for each local area and uses these local contrast values to determine weighing factors. This local quality approach ensures that regions with higher contrast contribute more to the similarity calculation, thereby improving extraction accuracy while maintaining manageable system complexity through localized processing.
2Manufacturing precision
If multiple stereo images are photographed under different photographing conditions, then the three-dimensional shape restoration can be improved, but the time and effort required for data collection increases
Solution Approach 1:
The patent performs preliminary analysis of local contrast values and weighing factor calculation before the actual corresponding point extraction. This preliminary action enables the system to optimize the use of existing single-pair stereo images, reducing the need for additional image capture under different conditions while maintaining high restoration precision.
Solution Approach 2:
The patent changes the parameter of contrast analysis by calculating local contrast values and using them to determine weighing factors. This parameter change allows the system to extract maximum information from a single pair of stereo images, eliminating the time-consuming process of capturing multiple images under different conditions while achieving precise three-dimensional shape restoration.
3Measurement precision
If all pixel information values are used equally in similarity calculation, then the calculation is simple, but the depth information lacks detail and precision
Solution Approach 1:
The patent calculates local contrast values for each local area and uses these to determine weighing factors. This local quality approach ensures that pixel information values in high-contrast regions are weighted more heavily, thereby enhancing depth information precision while keeping the processing complexity manageable through localized contrast analysis.
Solution Approach 2:
The patent performs preliminary calculation of contrast values and determination of weighing factors before the similarity calculation. This preliminary action prepares the pixel information values in advance, allowing the main similarity calculation to proceed efficiently with enhanced depth information precision without requiring complex real-time processing.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
There is provided an image analysis system which captures image data of an arbitrary pair of a first image RI and a second image LI among images obtained by color-photographing a single object from different positions into an analysis computer (10), wherein the computer includes corresponding point extraction means (30) for assigning a weighing factor to a pixel information value based on the contrast size of the pixel information value in each of a first local area ROI1 set around an arbitrary reference point in RI and second local areas ROI2s at which scanning is performed on LI, calculating the similarity between ROI1 and ROI2s, and extracting a corresponding point which corresponds to the reference point from a ROI2 having the highest similarity, and depth information calculating means (41) for calculating depth information of the object based on coordinates of the reference point and the corresponding point.