Planar Object Detection via Radial Logarithmic Image Transformation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for collecting and processing vertical road information, such as roadside signs, are time-consuming and require significant processing power due to the complexity of analyzing stereo pairs of images captured by mobile mapping vehicles.
Innovation Solution
The method involves transforming source images using a combined looking axis rotation and radial logarithmic space transformation to create images with identical viewing axes, allowing for the detection of planar objects using simple pattern matching algorithms, reducing the amount of data to be processed and identifying objects with similar dimensions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If stereo pair of images is used to capture position information, then position accuracy is improved, but processing complexity and processing power requirements increase
Solution Approach 1:
The patent applies preliminary action by performing looking axis rotation transformation and radial logarithmic space transformation on images before pattern matching. This preprocessing aligns the images in a standardized coordinate system and transforms them into a space where planar objects appear with consistent dimensions, thereby simplifying subsequent processing while maintaining position accuracy
Solution Approach 2:
The patent changes parameters by transforming images from standard Cartesian coordinate system to radial logarithmic space through the transformation formula: (u, v) = (log(r) * cos(θ), log(r) * sin(θ)), where r and θ are polar coordinates. This parameter transformation makes planar objects have identical dimensions across different viewing angles, enabling simpler pattern matching algorithms
2Measurement precision
If stereo pair of images is used to capture position information, then position accuracy is improved, but processing power requirements increase
Solution Approach 1:
The patent applies preliminary action by performing looking axis rotation transformation and radial logarithmic space transformation on images before pattern matching. This preprocessing aligns the images in a standardized coordinate system and transforms them into a space where planar objects appear with consistent dimensions, thereby simplifying subsequent processing while maintaining position accuracy
Solution Approach 2:
The patent extracts and processes only the relevant portions of images by transforming them into radial logarithmic space, where planar objects are highlighted with consistent characteristics. This extraction approach focuses computational power on identifying meaningful features rather than processing entire image datasets
3Measurement precision
If complex algorithms are used to analyze stereo pair of images, then position accuracy is improved, but processing time increases
Solution Approach 1:
The patent changes parameters by transforming images from standard Cartesian coordinate system to radial logarithmic space through the transformation formula: (u, v) = (log(r) * cos(θ), log(r) * sin(θ)), where r and θ are polar coordinates. This parameter transformation makes planar objects have identical dimensions across different viewing angles, enabling simpler pattern matching algorithms
Solution Approach 2:
The patent applies preliminary action by performing looking axis rotation transformation and radial logarithmic space transformation on images before pattern matching. This preprocessing aligns the images in a standardized coordinate system and transforms them into a space where planar objects appear with consistent dimensions, thereby simplifying subsequent processing while maintaining position accuracy
Data Source
Figure 1
Figure 2a~2b
Figure 3~6
AI summary
The invention relates to a method of identifying a planar object in source images. The method comprising, retrieving a first source image obtained by a first terrestrial based camera; retrieving a second source image obtained by a second terrestrial based camera; retrieving position data associated with the first and second source image; retrieving orientation data associated with the first and second source image; performing a looking axis rotation transformation on the first and second source image by means of the associated position data and orientation data to obtain first and second intermediate images, wherein the first and second intermediate images have an identical looking axis; - performing a radial logarithmic space transformation on the first and second intermediate images to obtain first and second radial logarithmic data images; detecting an area in the first image potentially being a planar object; comparing said potential planar object having similar dimensions in the second radial logarithmic data image and similar rgb characteristics; and finally, and identifying the area as a planar object and determining its position. The method enables the engineer to detect very efficiently planar perpendicular objects in subsequent images.