Region-Adaptive Motion Vectors for Optical Flow
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating high-quality motion vectors for optical flow processes is computationally intensive and inefficient, as conventional methods apply the same parameters to all image regions, including those with little movement, consuming excessive resources without improving overall image processing quality.
Innovation Solution
A processing system adjusts optical flow parameters based on the level of interest for each image region, reducing computations by generating higher-quality vectors for regions with significant movement and lower-quality vectors for regions with little movement, using rasterized motion vectors or visual features to determine interest levels.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional optical flow processes generate motion vectors by identifying matching pixels between input images, then motion vector quality is improved, but computational complexity and resource consumption increase significantly
Solution Approach 1:
The patent divides the image into multiple regions of interest (ROIs) and processes each region separately with different computational parameters. Regions with significant motion are identified and processed with full optical flow computation, while regions with little motion use simplified processing, thereby reducing overall computational complexity while maintaining motion vector quality where needed
Solution Approach 2:
The patent applies different levels of processing quality to different regions of the image based on their motion characteristics. High-quality motion vector computation is applied only to regions with significant motion, while low-quality or simplified processing is applied to regions with little motion, optimizing the balance between quality and computational cost
2Measurement precision
If optical flow processes are applied to all regions of an image, then motion detection accuracy is improved, but resource consumption and processing time increase
Solution Approach 1:
The patent performs a preliminary analysis of the image to identify regions of interest before applying full optical flow processing. By pre-identifying regions with significant motion characteristics, the system avoids applying computationally intensive processing to regions where it would be unnecessary, thereby improving processing efficiency while maintaining detection accuracy in relevant areas
Solution Approach 2:
The patent applies optical flow processing partially rather than uniformly across the entire image. By limiting full processing to only those regions where motion detection is actually needed (regions of interest), the system achieves adequate motion detection accuracy without the excessive resource consumption of processing every pixel
3Manufacturing precision
If high-quality motion vectors are generated for all image regions, then overall image processing quality is improved, but computational overhead increases significantly
Solution Approach 1:
The patent dynamically changes processing parameters based on region characteristics. For each region of interest, the system adjusts parameters such as search window size, number of iterations, and matching criteria according to the motion activity level. This allows high processing quality to be achieved in regions where it matters while reducing computational overhead in regions with little motion
Data Source
AI summary
A processing system performs a pre-pass of an optical flow process's input images to determine, for each region (e.g., each block) of an image, an associated level of interest. The level of interest for a region indicates the expected likelihood that an increased number of motion vector computations for that region will result in a higher quality output of an image processing pipeline. Accordingly, the processing system adjusts the parameters of the optical flow process for each region according to the region's corresponding level of interest, so that the optical flow process increases the number of motion vector computations for regions associated with a higher level of interest of interest and reduces the number of motion vector computations for regions associated with a lower level of interest.