Depth Map Generation via Vanishing Line Histograms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for generating depth maps based on vanishing points and lines are costly and not robust enough for efficient and effective direction detection and depth map generation.
Innovation Solution
A method that involves direction detection using the sum of absolute difference (SAD) along detected directions, calculation of pseudo-intercepts, generation of 2D and 1D histograms, selection of candidate vanishing lines, and weighted blending of global depth models to generate a robust and low-cost depth map.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current techniques for generating depth maps based on gradient planes and depth gradient assignment are used, then depth map generation is achieved, but the process is too costly and not robust enough for efficient direction detection
Solution Approach 1:
The patent segments the depth map generation process into distinct stages: gradient plane generation, depth gradient assignment, and consistency verification. By dividing the complex process into manageable segments, each can be optimized independently for both robustness and computational efficiency.
Solution Approach 2:
The patent performs preliminary direction detection and vanishing point identification before final depth map generation. By pre-processing the image to identify key geometric features and convergence points, the subsequent depth calculation becomes more efficient and robust.
2Measurement precision
If multiple steps including gradient planes generation and consistency verification are performed, then depth map accuracy is improved, but computational complexity and cost increase
Solution Approach 1:
The patent applies different processing qualities to different regions of the image. Gradient planes and depth gradients are calculated with higher precision in regions containing vanishing lines and convergence points, while other regions use simplified processing, optimizing the balance between accuracy and complexity.
Solution Approach 2:
The patent dynamically adjusts processing parameters based on local image characteristics. In regions with significant depth variations or vanishing point convergence, more sophisticated gradient analysis is applied, while uniform regions use simpler processing methods.
Data Source
AI summary
A method can include detecting a pixel direction and a pixel weight for each of a number of pixels in an image, generating one-dimensional (1D) and two-dimensional (2D) histograms based on the pixel directions and weights, and generating a global 2D histogram based on the generated 1D and 2D histograms. The method can also include generating a final depth map based on the global 2D histogram. The method can also include generating a block histogram statistic based on the pixel directions and pixel weights and checking the block histogram based on the block histogram statistic.