Image Generation Device Mesostructure Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image generation technologies fail to effectively produce high-quality images showing medium-level detail mesostructures, such as bumps on surfaces of objects like fruits, wood, or human skin, and cannot generate images under a pseudo light source, as they lack sufficient spatial resolution and detail in geometric parameters.
Innovation Solution
An image generation device that calculates geometric parameters from low-quality images using position information of light sources, viewpoints, and geometric normals, and modifies these parameters to enhance spatial resolution, allowing for the generation of high-quality images with mesostructures and pseudo light sources by applying these parameters to a reflection model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a low-quality image capturing device is used, then the device complexity and cost are reduced, but the image resolution and ability to show surface bumps deteriorate
Solution Approach 1:
The patent introduces an intermediary process between image capture and final output: geometric parameter estimation and mesostructure synthesis. The system estimates geometric parameters from the low-quality image and uses these as intermediaries to generate high-resolution output images, avoiding the need for high-quality capture hardware while achieving high resolution through computational processing
Solution Approach 2:
The patent replaces the mechanical/optical system (high-quality camera hardware) with a computational system. Instead of relying on physical resolution capabilities of the capturing device, the system uses image processing algorithms, geometric parameter estimation, and mesostructure synthesis to achieve high resolution, substituting mechanical capabilities with computational methods
2Device complexity
If conventional image generation methods are used, then the process is simple, but the ability to generate images under pseudo light sources and show mesostructures deteriorates
Solution Approach 1:
The patent changes the parameters used in image generation from simple pixel values to geometric parameters (surface normals, curvature, depth). By representing surfaces through geometric parameters rather than direct pixel data, the system enables flexible manipulation of lighting conditions and viewpoint, allowing pseudo light source generation and mesostructure visualization while maintaining a relatively straightforward processing pipeline
3Loss of information
If geometric parameters are estimated from image data, then the information regarding surface shape is obtained, but the spatial resolution and fineness of surface details deteriorate
Solution Approach 1:
The patent performs preliminary action by estimating geometric parameters (surface normals, curvature, depth) from the low-quality image before the actual high-resolution image generation. These pre-estimated geometric parameters serve as a foundation that guides the subsequent mesostructure synthesis process, allowing fine surface details to be generated based on the preliminary geometric information rather than requiring high-resolution input data
Solution Approach 2:
The patent transitions from two-dimensional image pixel data to three-dimensional geometric parameter representation. By estimating surface normals, curvature, and depth from 2D image data, the system creates a 3D geometric model that can then be used to synthesize mesostructures with high spatial resolution, effectively adding dimensional information that compensates for the low resolution of the original image
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
The device generates high-quality images clearly showing surface details like bumps and allows for the simulation of images under desired pseudo light sources, improving image resolution and realism, particularly suitable for digital cameras and security systems.
Implementation Method 1
applying the modified geometric parameter to a reflection model for deciding the pixel value
Data Source
AI summary
Provided is an image generation device generating a high-quality image of an object under a pseudo light source at any desired position, based on geometric parameters generated from a low-quality image of the object. The image generation device includes: a geometric parameter calculation unit that calculates a first geometric parameter regarding a shape of a surface from light source position, viewpoint position, and geometric normal information regarding the shape; a high-resolution database unit that stores an exemplum indicating a mesostructure of a portion of the surface and has a spatial resolution higher than the geometric normal information; an exemplum expansion unit that increases the exempla to be spatially expanded; a geometric parameter modification unit that modifies the first geometric parameter using the increased exempla; and an image generation unit that generates an output image by applying the modified geometric parameter to a reflection model.