Planar Contour Recognition Through Temporal Edge-Point Fitting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Deep-learning-based methods for recognizing planar contours of target objects are time-consuming due to reliance on pre-training, affecting efficiency in applications like robot sorting and object manipulation.
Innovation Solution
A method involving fitting edge points from target and previous frame images to construct a fitting graph, deleting outlier edge points, and recognizing the planar contour from remaining points, reducing the need for deep learning and improving recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep-learning-based method is used for planar contour recognition, then recognition accuracy is improved, but training time is excessive and efficiency is reduced
Solution Approach 1:
The patent extracts and utilizes pre-existing edge point data from video frames to construct fitting graphs for contour recognition, eliminating the need for extensive pre-training. By extracting edge points and fitting them to form contours directly from operational data, the system achieves accurate recognition without time-consuming training processes.
Solution Approach 2:
The system performs self-service by automatically generating training data from its own operational video frames. The edge points and fitting graphs are created during normal operation without requiring external training datasets, allowing the system to improve its recognition capabilities autonomously during use rather than requiring separate training phases.
2Measurement precision
If deep-learning-based method is used for planar contour recognition, then recognition accuracy is improved, but computational complexity and processing time are increased
Solution Approach 1:
The patent segments the contour recognition task into discrete steps: extracting edge points, fitting curves to edge points, forming closing contours, and deleting unnecessary points. This segmentation breaks down the complex recognition process into manageable operations that can be performed sequentially with simpler computations rather than requiring complex deep learning models.
Solution Approach 2:
The patent replaces the mechanical deep learning system with a geometric fitting approach. Instead of using neural networks to recognize contours, the system uses mathematical curve fitting algorithms to construct contours from edge points, significantly reducing computational complexity while maintaining recognition accuracy.
3Measurement precision
If edge points from multiple frames are fitted to construct contour, then recognition accuracy is improved, but data processing volume is increased
Solution Approach 1:
The patent performs preliminary actions by extracting and storing edge point data from multiple video frames in advance. These pre-extracted edge points are then reused in the fitting process, avoiding the need to reprocess the entire video data for each contour recognition task and reducing the effective data processing volume.
Solution Approach 2:
The patent merges edge point data from multiple frames into a single fitting graph that represents the complete contour. By combining information from multiple sources (different frames) into one integrated representation, the system achieves accurate recognition without processing each frame separately, thus reducing overall processing volume.
Data Source
AI summary
This application relates to a planar contour recognition method and apparatus, a computer device, and a storage medium. The method includes obtaining a target frame image collected from a target environment; fitting edge points of an object plane in the target frame image and edge points of a corresponding object plane in a previous frame image to obtain a fitting graph, the previous frame image being collected from the target environment before the target frame image; deleting edge points that do not appear on the object plane of the previous frame image, in the fitting graph; and recognizing a contour constructed by remaining edge points in the fitting graph as a planar contour.