Event Vision Sensor Frame Overlap for High-Speed Object Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The accuracy of inspection or recognition of objects using event-based vision sensors (EVS) can be lowered when the inspection target object moves at a higher speed than expected or when multiple targets are intermingled.
Innovation Solution
An information processing apparatus that generates image data based on event data from a first sensor with a pixel array unit, where the processing unit creates first image data of a current frame by using event data detected during an integration period that includes an overlap period with at least part of the frame period of a previous frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If event data is collected during a standard frame period without overlap, then the processing speed and frame rate are maintained, but the accuracy of inspection and recognition of high-speed moving objects deteriorates
Solution Approach 1:
The patent applies preliminary action by extending the integration period to include an overlap with the previous frame period. This allows the system to capture event data from high-speed moving objects that may have moved between standard frame intervals, ensuring these objects are included in the current frame's image data generation, thus improving detection accuracy without requiring complex real-time adjustments during frame processing
Solution Approach 2:
The patent implements dynamics by making the integration period flexible and adaptable. Instead of using a fixed frame period for data collection, the system dynamically extends the integration period to overlap with the previous frame, allowing the integration window to adapt to the speed and movement patterns of objects being inspected, thereby maintaining high measurement precision for high-speed objects
2Measurement precision
If the integration period is extended to include overlap with previous frame, then the detection accuracy of high-speed moving objects is improved, but the processing time and data volume increase
Solution Approach 1:
The patent applies merging by combining event data from the overlap period with data from the non-overlap period within the same integration period. This allows the system to process extended integration data efficiently by treating the overlap portion as supplementary information that can be integrated with the primary frame data, reducing the overall processing burden compared to handling completely separate frame periods
Solution Approach 2:
The patent uses partial action by selectively processing event data only from the extended integration period when high-speed objects are detected or anticipated. Instead of continuously processing full overlap periods for all frames, the system applies the extended integration approach selectively, thereby improving detection accuracy for high-speed objects while minimizing the overall processing time and computational load
3Measurement precision
If event data from overlap period is used for current frame, then tracking accuracy of multiple intermingled targets is improved, but the data processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the event data processing into distinct temporal segments: the overlap period portion and the non-overlap period portion. This segmentation allows the system to separately identify and process events from different time windows, making it easier to track multiple intermingled targets by associating events with their respective temporal and spatial contexts, thereby reducing processing complexity while maintaining high tracking accuracy
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach helps to suppress the decrease in accuracy of object inspection, recognition, and tracking by ensuring that image data is generated with an overlap period that accounts for the movement of objects, thereby improving the detection and recognition of high-speed moving objects and objects with intersecting trajectories.
Implementation Method 1
a pixel array unit in which a plurality of pixels that each detect a luminance change of incident light and output event data are arranged
Data Source
AI summary
The present invention aims to suppress a decrease in the accuracy of inspection, recognition, and the like of an object. An information processing apparatus according to embodiments includes a processing unit (202) that generates image data on the basis of event data output from a first sensor (100) including a pixel array unit (101) in which multiple pixels (20) that each detect a luminance change of incident light and output event data are arranged in a two-dimensional lattice pattern. In the information processing apparatus, the processing unit generates first image data of a current frame by using event data detected during an integration period including an overlap period that overlaps with at least a part of a frame period of a previous frame.