Adaptive Interpolation for De-interleaved 3D View Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies face challenges in effectively delivering and up-converting stereoscopic video, particularly in checkerboard interleaved formats, which results in suboptimal quality and increased complexity during real-time decoding and display.
Innovation Solution
The method involves receiving a data stream with interleaved views, determining disparity between views using filters, and applying content-adaptive interpolation techniques to up-convert pixel values, utilizing disparity analysis and multi-hypothesis filtering to improve image quality while maintaining or reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If checkerboard interleaved format is used for delivering stereoscopic video, then delivery efficiency and bandwidth utilization are improved, but image quality and processing complexity deteriorate
Solution Approach 1:
The patent applies segmentation by dividing the image into multiple blocks or regions (e.g., 8x8 pixel blocks) and processing each segment independently. This allows the system to maintain the efficiency benefits of checkerboard interleaving while applying localized interpolation and filtering operations to restore image quality in each segment, thereby resolving the contradiction between delivery efficiency and image quality.
Solution Approach 2:
The patent implements local quality by applying different interpolation and filtering operations to different regions of the image based on their specific characteristics. Regions with high disparity or edge content receive different processing than smooth regions, allowing the system to optimize image quality locally while maintaining overall delivery efficiency of the interleaved format.
2Device complexity
If traditional fixed interpolation methods are used for up-converting de-interleaved views, then processing simplicity is maintained, but interpolation accuracy and adaptability deteriorate
Solution Approach 1:
The patent applies dynamics by transitioning from fixed interpolation methods to adaptive interpolation that dynamically adjusts parameters based on local image characteristics. The system calculates disparity maps and selects appropriate interpolation filters (e.g., bilinear, bicubic, or edge-preserving filters) based on the specific content of each region, thereby improving interpolation accuracy without requiring overly complex processing.
Solution Approach 2:
The patent implements parameter changes by varying interpolation parameters such as filter type, kernel size, and strength based on local disparity magnitude and image content. This allows the system to optimize interpolation accuracy for different regions while keeping the overall processing framework relatively simple by using a finite set of predefined interpolation modes that are selected based on content analysis.
3Manufacturing precision
If high-quality interpolation filters are applied to all regions uniformly, then image quality is improved, but processing time and computational complexity increase
Solution Approach 1:
The patent applies local quality by classifying image regions based on disparity magnitude and content characteristics, then applying appropriate interpolation filters selectively. High-quality filters are applied only to regions where they are most needed (e.g., low disparity regions or regions with smooth content), while simpler filters are used for high disparity or edge regions, thereby maintaining image quality where critical while reducing overall processing time.
Solution Approach 2:
The patent implements partial action by applying high-quality interpolation only to specific portions of the image rather than uniformly to all regions. The system identifies and processes only the necessary portions (e.g., regions with low disparity or smooth content) using computationally intensive filters, while using faster, simpler methods for other regions, thereby achieving good average image quality with reduced processing time.
Data Source
AI summary
Method for reconstructing interleaved views of a 3D image comprising the steps of: receiving interleaved views of a 3D image, de-interleaving said interleaved views into a first view and a second view, each de-interleaved view comprising missing pixels, at each pixel location determining an amount of disparity between said first and second views, for each view interpolating each missing pixel based on the amount of disparity determined at the associated pixel location, wherein when the determined disparity is above a predetermined disparity threshold, the missing pixel value is interpolated from the data of the view to be interpolated only (i.e. the view comprising said missing pixel value), and when the determined disparity is at or below said predetermined disparity threshold, the missing pixel value is interpolated from the data of both views.