Transparent Object Detection Using Polarization Surface Normal Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image detection techniques struggle to stably detect objects with high light transmittance, such as colorless glass or acrylic resin, due to limited color information and susceptibility to surrounding light conditions, leading to detection failures or misidentification.
Innovation Solution
An information processing apparatus and method that acquire polarization images in multiple orientations, derive normal vectors, assume the presence of undetected surfaces, and confirm the presence of the subject using these vectors to generate definitive surface data, enabling accurate detection of transparent objects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional image detection techniques are used, then detection process is simple, but detection reliability fails for objects with high light transmittance
Solution Approach 1:
The patent changes the detection parameter from conventional intensity-based imaging to polarization-based imaging. By acquiring polarization images in multiple orientations and analyzing polarization characteristics, the system can reliably detect transparent objects that are invisible or difficult to detect with conventional imaging methods.
Solution Approach 2:
The patent introduces polarization information as an intermediary to enable detection of transparent objects. The polarization images serve as an intermediate representation that contains information about the optical properties of transparent surfaces, which is then used to improve detection reliability.
2Measurement precision
If polarization images in multiple orientations are acquired, then detection precision improves, but measurement complexity increases
Solution Approach 1:
The patent segments the polarization measurement into discrete orientations. By acquiring polarization images at specific angular intervals (e.g., 0°, 45°, 90°, 135°), the system achieves precise surface detection while managing measurement complexity through structured sampling rather than continuous measurement.
Solution Approach 2:
The patent uses a sufficient number of polarization orientations to achieve accurate surface normal estimation without requiring exhaustive measurement. Typically 3-4 orientations provide enough information to reliably determine surface properties while avoiding unnecessary measurement complexity.
3Reliability
If surface continuity is assumed to fill undetected regions, then detection completeness improves, but risk of false detection increases
Solution Approach 1:
The patent performs preliminary surface detection using polarization images before attempting to fill undetected regions. By first establishing reliable detected surfaces with high confidence, the system can then make informed assumptions about continuity in ambiguous regions, reducing false detections while improving completeness.
Solution Approach 2:
The patent uses feedback from the detected surface normals and polarization characteristics to guide the surface completion process. The estimated normal vectors from detected regions provide feedback that constrains the assumption of continuity, ensuring that filled surfaces are consistent with the observed optical properties and reducing false detections.
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 allows for stable detection of objects with high light transmittance by utilizing polarization characteristics, improving detection accuracy and reducing misidentification, even in challenging lighting conditions.
Implementation Method 1
acquire, from an imaging apparatus, data of polarization images in a plurality of orientations
Data Source
AI summary
An image acquisition section of an information processing apparatus acquires, from an imaging apparatus, polarization images in a plurality of orientations. A normal acquisition section of an image analysis section detects a surface of a subject by acquiring a normal vector on the basis of orientation dependence of polarization luminance. A surface assumption section assumes the presence of an undetected surface continuing from the detected surface. A subject detection section confirms whether or not the assumed surface is present using a normal vector estimated for the assumed surface. An output data generation section generates and outputs output data using information regarding the detected surface.