Strip-like Line Segment Detection in High-Resolution Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image detection methods fail to accurately characterize target structures in high-resolution images, as they ignore width information of line segments, leading to ineffective feature representation.
Innovation Solution
An image detection method that establishes a data model using mark point information of strip-like line segments, including length, width, direction, and center coordinates, and a priori model based on the distribution and number of line segments, employing probability density functions and sampling techniques to obtain a globally optimal solution for target model detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If width information of line segments is ignored to simplify the detection model, then the model construction is easier and faster, but the detection accuracy deteriorates in high-resolution images
Solution Approach 1:
The patent transforms the line segment representation from 1D (ignoring width) to 2D strip-like structures by introducing width as a new parameter. This parameter change enables the model to accurately represent target structures in high-resolution images while maintaining computational feasibility through probabilistic modeling approaches.
Solution Approach 2:
The patent adds the width dimension to line segments, transforming them from simple 1D elements into 2D strip-like structures. This dimensional expansion allows the model to capture the full geometric characteristics of target structures, resolving the contradiction between model simplicity and detection accuracy.
2Measurement precision
If strip-like line segments with width information are used to improve feature representation, then detection accuracy improves, but the data model complexity increases
Solution Approach 1:
The patent replaces traditional deterministic geometric modeling with probabilistic modeling using point process theory. This substitution allows the complex strip-like line segment data to be handled through probability density functions and statistical methods, simplifying the overall model while maintaining high detection accuracy.
Solution Approach 2:
The patent introduces probability density functions as an intermediary layer between the raw strip-like line segment data and the detection algorithm. This intermediary transforms complex geometric information into probabilistic representations, making the data more manageable and the model less complex while preserving feature representation accuracy.
3Productivity
If traditional line segment models are used for low-resolution images, then the detection process is efficient, but the model fails to characterize target structures accurately in high-resolution images
Solution Approach 1:
The patent creates a dynamic modeling approach where the line segment representation adapts to image resolution requirements. For low-resolution images, the model can operate with simplified representations, while for high-resolution images, it automatically utilizes the full strip-like structure with width information, maintaining both efficiency and reliability across different scenarios.
Data Source
AI summary
Embodiments of the present disclosure disclose an image detection method including: establishing a data model based on mark point information of a group of first strip-like line segments, the mark point information of each first strip-like line segment including the length, the width, the direction and the coordinate of the center point of the first strip-like line segment, establishing a priori model based on the distribution and the number of a group of second strip-like line segments which constitute a target structure, establishing a probability density function for the data model and the priori model, and performing sampling and solution optimization to obtain a globally optimal solution, and detecting the image taking the globally optimal solution as a target model. The embodiments of the present disclosure also disclose an image detection system and a non-volatile computer readable medium.