Image Processing Apparatus Computational Parameter Synthesis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing methods require a large number of photographic images when multiple parameters are changed, making the process impractical for generating images with different values of multiple photographic parameters.
Innovation Solution
An image processing apparatus and method that generates images with different values of multiple photographic parameters by learning from a sparse set of images and predicting densely sampled images using a combination of focus position and f-stop values, employing a learning apparatus with vector generation, normalization, and clustering to determine pixel values and generate predicted images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple photographic parameters are changed to generate diverse images, then image processing versatility is improved, but the number of required photographic images increases exponentially
Solution Approach 1:
The patent performs preliminary actions by capturing a limited set of photographic images at different focus positions before the actual processing stage. These pre-captured images serve as a foundation stack that can be computationally transformed to generate multiple virtual images with different parameters, avoiding the need to capture all possible parameter combinations in advance.
Solution Approach 2:
The patent creates virtual copies of photographic images through computational processing. By using the captured image stack and applying algorithms for refocusing, depth-of-field control, and viewpoint conversion, the system generates multiple synthesized images that replicate the effect of capturing images with different parameters, thereby reducing the need for actual multiple physical captures.
2Manufacturing precision
If a large number of photographic images are captured to achieve high-quality processing results, then image quality is improved, but the time and effort required for photographing increases
Solution Approach 1:
The patent changes the approach from varying multiple parameters simultaneously to varying only the focus position parameter while keeping other parameters constant. This parameter simplification allows high-quality image processing through computational methods rather than requiring extensive physical captures of multiple parameter combinations, significantly reducing photographing time while maintaining or improving image quality.
Solution Approach 2:
The patent replaces the mechanical system of capturing multiple physical photographs with different parameters with a computational system that processes a single captured stack to generate virtual images. This substitution of mechanical capture with computational generation reduces the time and effort required for photographing while achieving comparable or superior results.
3Adaptability or versatility
If multiple photographic parameters are varied to achieve comprehensive image coverage, then processing completeness is improved, but device complexity increases
Solution Approach 1:
The patent makes the photographing system universal by designing it to capture images with a fixed set of parameters that can serve multiple processing purposes. The same captured image stack can be used for refocusing, depth-of-field control, viewpoint conversion, and other processing tasks, eliminating the need for specialized equipment for each function and reducing overall system complexity.
Data Source
AI summary
Provided is an image processing apparatus including an image generation unit that, from photographic images that are captured using multiple photographic parameters, generates an image of which values of the multiple photographic parameters are different from values of the photographic image.