Normal Line Calculation Using Polarized Image Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image analysis techniques for acquiring the position and posture of a target object from captured images face instability due to the object's appearance, position, and imaging environment, leading to inaccurate processing, especially when feature points are scarce or the object is small and far away.
Innovation Solution
An information processing apparatus that divides the image plane into regions and assigns either a specular reflection model or a diffuse reflection model to each region, calculating and integrating the normal line distributions to generate a comprehensive normal line distribution for accurate object state acquisition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If feature point-based matching techniques are used to extract target object images, then the processing can be performed with simpler algorithms, but the accuracy deteriorates when feature points are scarce or the object is small and far away
Solution Approach 1:
The patent changes the parameter being measured from feature point coordinates to polarization angle distribution characteristics. By using polarization parameters instead of feature point parameters, the system achieves accurate target identification even when feature points are scarce or the object is small and distant, without requiring complex feature matching algorithms
Solution Approach 2:
The patent substitutes geometric feature-based matching with polarization-optical property-based analysis. Instead of relying on mechanical/geometric feature point detection and matching, the system uses polarization angle distribution which provides intrinsic optical characteristics of the target, achieving both simplicity and accuracy
2Measurement precision
If spatial and temporal granularity are refined or more complex algorithms are applied to improve processing accuracy, then the measurement precision improves, but the processing load increases
Solution Approach 1:
The patent changes from analyzing spatial feature point distributions to analyzing polarization angle distributions. This parameter transformation allows accurate target characterization without requiring fine spatial granularity or complex algorithms, thereby maintaining high processing accuracy while reducing computational load and improving processing efficiency
3Productivity
If a single reflection model is applied to the entire image, then the processing is simpler and faster, but the accuracy deteriorates when both specular and diffuse reflection regions are present
Solution Approach 1:
The patent segments the image processing by applying different reflection models to different regions. The region division section divides the image into specular reflection regions and diffuse reflection regions, then applies appropriate models to each region separately. This segmentation enables accurate normal line calculation across the entire image while maintaining reasonable processing speed
Solution Approach 2:
The patent implements local quality by assigning different reflection model characteristics to different image regions based on their local properties. Specular regions use specular reflection models while diffuse regions use diffuse reflection models, allowing each region to be processed with the most appropriate method for its local characteristics, thereby achieving high accuracy without excessive processing load
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables efficient and accurate acquisition of the target object's state using captured images, improving processing accuracy without the need for complex algorithms or heavy processing loads.
Implementation Method 1
acquires a polarized image of the target space
Implementation Method 2
obtain a distribution of normal vectors of a specular reflection region through application of a specular reflection model
Implementation Method 3
obtain a distribution of normal vectors of a diffuse reflection region through application of a diffuse reflection model
Data Source
AI summary
A captured image acquisition section 50 acquires from an imaging apparatus 12 data of a polarized image captured of a target object and stores the image data into an image data storage section 52. A region division section 58 in a normal line information acquisition section 54 divides a plane of the image into regions according to a predetermined criterion such as a polarization degree or luminance. A normal line calculation section 60 obtains a distribution of normal lines of each region by applying either a specular reflection model or a diffuse reflection model thereto. An integration section 62 integrates the distributions of normal lines of the regions into a normal line distribution of the entire image. An output data generation section 56 performs information processing using the normal line distribution and outputs the result of the processing.