Pseudo 3D Image Creation Using Statistical Pixel Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing pseudo 3D image creation methods fail to consistently provide a strong 3D perception to viewers, as they tend to yield varying degrees of 3D effect based on the complexity of patterns and edges in non-3D images, with simple patterns and few edges resulting in a weak 3D feeling.
Innovation Solution
A pseudo 3D image creation apparatus that calculates statistical pixel values in non-3D images to generate evaluation values, combines basic depth models based on these values, and shifts the texture of the image to create a different-viewpoint signal, effectively enhancing the 3D perception by generating a stereo pair for display.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the device uses high-frequency component values to determine composition ratio of basic depth models, then the 3D effect is enhanced for images with complicated patterns and edges, but the 3D effect becomes weak for images with simple patterns and few edges
Solution Approach 1:
The invention changes the parameter used for composition ratio determination from high-frequency component values to statistical amounts of pixel values in predetermined areas. This parameter change enables the system to adapt to both complicated and simple image patterns effectively, resolving the contradiction between enhancing 3D effect for complex images and maintaining adaptability for simple images.
Solution Approach 2:
The invention introduces dynamic adjustment of composition ratio based on statistical analysis of pixel values in different regions. The system dynamically selects appropriate depth models and composition ratios adaptively, allowing the pseudo-3D image creation to work consistently across various image types without requiring manual intervention or fixed parameters.
2Measurement precision
If the device combines basic depth models based on high-frequency component values, then the depth estimation is improved for complex scenes, but the depth estimation becomes inaccurate for simple scenes
Solution Approach 1:
The invention changes the parameter from high-frequency component values to statistical amounts of pixel values in predetermined areas. This enables more accurate depth estimation for simple scenes by using statistical distribution information rather than edge-based high-frequency analysis, thereby improving both precision and reliability across different scene types.
Solution Approach 2:
The invention segments the image into predetermined areas and calculates statistical amounts of pixel values for each segment. This segmentation approach allows the system to analyze different regions of the image independently, improving depth estimation accuracy by considering local statistical characteristics rather than relying solely on global high-frequency content.
Data Source
AI summary
A plurality of basic depth models indicate depth values of a plurality of basic scene structures. Statistical amounts of pixel values in predetermined areas in a non-3D image are calculated to generate first evaluation values. A statistical amount of pixel values in a whole of the non-3D image is calculated to generate a second evaluation value. The basic depth models are combined into a combination result according to a combination ratio depending on the generated first evaluation values. Depth estimation data is generated from the combination result and the non-3D image. A texture of the non-3D image is shifted in response to the generated depth estimation data and the generated second evaluation value to generate a different-viewpoint picture signal. The generated different-viewpoint picture signal and a picture signal representative of the non-3D image make a stereo pair representing a pseudo 3D image.