Polarized Image Processing for Shadow Distinction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image analysis for recognizing the position or posture of real objects from photographed images is prone to inaccuracies due to environmental factors such as illuminance and surrounding objects, making it difficult to distinguish real objects from shadows.
Innovation Solution
An information processing apparatus and method that utilize polarized image processing to acquire distributions of degrees of polarization and normal vectors, allowing for the specification of shadow areas and subsequent processing based on these characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional image analysis is used to recognize position or posture of real objects, then various events in the real world can be taken as input data, but accuracy of processing is easily influenced by circumferential situation such as other things or illuminance
Solution Approach 1:
The patent introduces shadow maps as an intermediary data structure that mediates between the photographed image and the object recognition process. The shadow map pre-calculates and stores shadow information based on object positions and light source locations, serving as a reference that helps distinguish real objects from shadows in the photographed image, thereby improving recognition accuracy without limiting versatility
Solution Approach 2:
The patent changes the parameter representation by introducing multiple parameters including shadow map data, light source position, and object position information. By representing shadow regions through these additional parameters rather than relying solely on image intensity, the system can accurately differentiate between actual objects and shadow regions even under varying illuminance conditions
2Adaptability or versatility
If image analysis is performed under varying illuminance conditions, then various lighting scenarios can be handled, but it becomes difficult to distinguish real objects from shadows
Solution Approach 1:
The shadow map acts as an intermediary that decouples the relationship between illuminance variations and object-shadow distinction. By pre-calculating shadow regions based on geometric relationships between objects and light sources, the system creates a reference that remains valid across different lighting scenarios, enabling accurate distinction without being affected by illuminance changes
Solution Approach 2:
The system performs preliminary calculation of shadow maps before actual object recognition. By pre-determining where shadows should fall based on object positions and light source locations, the system prepares reference information in advance that enables rapid and accurate distinction between objects and shadows during real-time processing, regardless of lighting conditions
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 enhances the accuracy of recognizing the state of a real space and diversifies information processing by effectively distinguishing between real objects and shadows, even when they are out of the field of view or under varying light conditions.
Implementation Method 1
a polarized image processing section which acquires distributions of degrees of polarization and normal vectors in an image plane by using a photographed polarized image
Data Source
AI summary
A photographed image acquiring section 50 acquires data associated with a polarized image obtained through photographing of an object from an imaging apparatus 12, and stores the data in an image data storing section 52. A polarized image processing section 58 of a real space recognizing section 54 obtains distributions of degrees of polarization or normal vectors in an image plane by using the polarized image. A shadow area specifying section 60 specifies an area having peculiarity in the distribution of the degrees of polarization or the normal vectors in an image representing one surface of a subject as an area of a shadow. A state specifying section 62 acquires information associated with a state of a real object or the shadow in the real space on the basis of the area of the image of the shadow. An output data generating section 56 executes information processing by using state information in the real space, and outputs a processing result.