Fisheye Learning Image Generation With Virtual Road Markings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies for detecting the connection state between a tow vehicle and a towed vehicle using image processing require large amounts of learning data, necessitating the actual capture of fisheye images with road surface paint, which is inefficient and resource-intensive.
Innovation Solution
A method to generate a learning fisheye image by applying planar orthogonalization transformation to an original fisheye image and adding virtual road surface paint, enabling the creation of learning data without physically capturing such images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If actual fisheye images with road surface paint are captured using a learning camera, then the model learning accuracy is improved, but the resource consumption and time required increase significantly
Solution Approach 1:
The patent creates virtual copies of road surface paint markings by generating synthetic fisheye images that include virtual road surface paint. Instead of capturing actual images with physical paint markings, the system generates digital representations of these markings through image processing and virtual element addition, thereby eliminating the need for time-consuming physical setup and capture while maintaining model learning accuracy
Solution Approach 2:
The patent performs preliminary processing by pre-generating virtual road surface paint elements and pre-computing their positions and appearances in the fisheye image coordinate system. This preliminary preparation allows the learning model to be trained with pre-prepared synthetic images containing virtual paint markings, avoiding the need for actual field capture operations during the learning data preparation phase
2Measurement precision
If actual fisheye images with road surface paint are captured using a learning camera, then the learning data quality is improved, but the cost and complexity of data collection increase
Solution Approach 1:
The patent replaces the mechanical/physical system of actual paint application and image capture with a computational system. Instead of physically painting road surfaces and using cameras to capture images, the system uses image processing algorithms to generate virtual paint markings and合成 fisheye images computationally, thereby simplifying the data collection system while maintaining learning data quality
Solution Approach 2:
The patent introduces virtual road surface paint as an intermediary element between the physical road and the learning model. These virtual paint elements serve as mediators that carry the necessary visual information for model training without requiring actual physical paint application, thus reducing the complexity of the data collection process while preserving learning data quality
3Productivity
If fisheye images are transformed to planar images and virtual road surface paint is added, then the learning data generation efficiency is improved, but the transformation process complexity increases
Solution Approach 1:
The patent segments the image processing task into distinct stages: first transforming the fisheye image to a planar orthogonalization transformation image, then separately generating and adding virtual road surface paint elements to the transformed image. This segmentation allows each processing step to be optimized independently and facilitates efficient batch processing of multiple images for learning data generation
Solution Approach 2:
The patent utilizes parameter transformations associated with the fisheye to planar orthogonalization transformation to efficiently generate corresponding virtual paint positions and shapes. By changing the coordinate system parameters and applying the transformation matrix, the system automatically adjusts virtual paint element parameters to match the transformed image geometry, enabling efficient learning data generation without manual intervention
Data Source
AI summary
A learning image generation device generates a learning fisheye image for use in learning of a model used for an estimation of a state of a trailer based on a fisheye image shot by a camera mounted on a vehicle towing the trailer via a tow bar, acquires an original learning fisheye image shot by a learning camera mounted on a learning vehicle towing a learning trailer via a learning tow bar, generates a planar orthogonalization transformation image by executing planar orthogonalization transformation on the original learning fisheye image, adds virtual road surface paint to the planar orthogonalization transformation image, and generates the learning fisheye image by executing inverse transformation of the planar orthogonalization transformation on the planar orthogonalization transformation image after the virtual road surface paint is added (planar image having the virtual road surface paint).