Image Processing for High-Speed 3D Multi-Object Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The exponential increase in calculation amount due to the number of vertices in 3D model tracking processing makes it difficult to complete the processing within a desired time as the number of target objects increases.
Innovation Solution
An image processing apparatus that obtains time-series shape data, conducts tracking processing on each object based on correspondence relation information, and outputs tracked data with metadata associating object identification and track information, allowing high-speed processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If tracking processing is conducted on a large number of objects in mesh format, then tracking accuracy is maintained, but calculation amount exponentially increases
Solution Approach 1:
The patent segments the tracking processing by introducing keyframe concepts and dividing the frame sequence into multiple tracks. Each track independently tracks objects between keyframes, breaking down the overwhelming calculation task into manageable segments. This segmentation allows the system to maintain tracking accuracy while reducing the exponential calculation burden by processing smaller frame intervals separately.
Solution Approach 2:
The patent performs preliminary actions by pre-identifying keyframes and pre-establishing track structures before conducting detailed vertex tracking. Correspondence relation information is prepared in advance for keyframes, and object identification is established beforehand. This preliminary organization enables the tracking algorithm to work with pre-structured data, significantly reducing real-time calculation requirements while maintaining accuracy.
2Quantity of substance
If the number of target objects for tracking processing increases, then tracking coverage is improved, but processing time exceeds desired time
Solution Approach 1:
The patent applies segmentation by dividing the processing of multiple objects into independent tracks, each handling a subset of objects between keyframes. This allows parallel processing of different tracks and reduces the time complexity from exponential to manageable levels. The system can track a large number of objects by distributing the computational load across multiple segmented tracks rather than processing all objects simultaneously in a single exhaustive analysis.
Solution Approach 2:
The patent changes parameters by introducing keyframe intervals and track-specific processing parameters. Instead of uniformly processing all frames for all objects, the system adjusts processing parameters based on keyframe distances and object importance. This parameter adaptation enables efficient processing of large numbers of objects by focusing computational resources on critical tracking intervals and less frequently updated objects.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
High-speed tracking processing on a large number of objects is achieved. Time-series shape data composed of a frame group in which each of frames contains 3D models representing three-dimensional shapes of a plurality of objects, respectively, is obtained. Then, tracking processing is conducted for each object based on correspondence relation information contained in the obtained time-series shape data.