Boundary Extraction Method for Non-Contact Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional boundary extraction methods in non-contact imaging devices, such as top-mounted scanners, face challenges in accurately extracting target boundaries due to interference from noise boundaries, leading to reduced accuracy in image correction and page boundary detection.
Innovation Solution
A method and apparatus that enhance the gradient of target boundaries while weakening noise boundaries by adjusting gradients based on differences between adjacent regions, followed by iterative detection and correction to improve boundary tracking accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If gradient-based boundary tracking is used, then boundary extraction can be performed, but noise boundaries from other pages interfere with target boundary detection accuracy
Solution Approach 1:
The patent applies local quality by differentiating gradient characteristics between target and noise boundaries. It calculates gradient magnitudes and directions locally at each pixel, then uses statistical analysis (mean, standard deviation) of gradient values in neighboring regions to identify whether a boundary is a target boundary (current page) or noise boundary (other pages). This local analysis enables selective enhancement of target boundaries while suppressing noise boundaries.
Solution Approach 2:
The patent changes gradient parameters adaptively based on local boundary characteristics. It computes gradient magnitudes and directions, then modifies these gradient values by adding enhancement terms for target boundaries and suppression terms for noise boundaries. The enhancement term increases gradient magnitude for target boundaries, while the suppression term decreases gradient magnitude for noise boundaries, thereby changing the gradient parameter distribution to improve boundary tracking accuracy.
2Adaptability or versatility
If traditional gradient tracking is applied to thick objects, then imaging is possible, but curved surface deformations occur due to imaging limitations
Solution Approach 1:
The patent performs preliminary boundary extraction and correction operations before final image processing. It first extracts boundaries from the raw image containing curved surface deformations, then uses these extracted boundaries to correct the geometric distortions. By performing boundary extraction first and using it to guide subsequent correction operations, the method eliminates the need for physical separation of pages while maintaining high geometric accuracy.
3Measurement precision
If gradient enhancement is applied to all boundaries, then boundary detection sensitivity increases, but noise boundaries are also enhanced leading to tracking errors
Solution Approach 1:
The patent implements feedback by using the extracted boundary information to guide gradient enhancement. It first performs initial boundary extraction, then uses the extracted boundary positions and characteristics to determine which gradients should be enhanced and which should be suppressed. The feedback loop continuously refines boundary detection by comparing extracted boundaries with gradient information, adjusting gradient enhancement parameters based on the reliability of detected boundaries.
Data Source
AI summary
The present invention discloses a boundary extraction method and apparatus, the method including: a gradient estimation step of estimating a gradient of each pixel in a captured image; a gradient adjustment step of adjusting, by enhancing a gradient of a target boundary of an object contained in the captured image and weakening a gradient of a noise boundary, the estimated gradient, so that the adjusted gradient is considered as a current gradient; and a boundary extraction step of extracting a boundary of the object based on the current gradient. According to the embodiments of the invention, in a case of using a non-contact imaging device to capture an image, it is possible to more accurately extract a boundary of an object contained in the captured image.