Moving Object Detection via Spatiotemporal Inter-Leg Periodicity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing moving object detection technologies face challenges in accurately detecting objects with two or more legs, such as humans, due to issues like mistaken detection, influence of shadows and lighting changes, and instability in parameterization caused by occlusion and environmental fluctuations.
Innovation Solution
A moving object detection device that generates spatiotemporal data from sensor outputs, extracts inter-leg information, and analyzes periodicity to determine the presence and movement of objects with two or more legs, using units like spatiotemporal data generation, inter-leg information extraction, periodicity analysis, and movement information generation to improve detection precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If still image or single differential image is used for moving object detection, then detection speed is fast, but detection accuracy deteriorates due to mistaken detection and inability to eliminate shadow influence
Solution Approach 1:
The patent applies preliminary action by performing background subtraction to generate a differential image before further processing. This preliminary step removes static background elements, allowing the system to focus on moving objects while maintaining fast processing speed. The differential image serves as a pre-processed input that accelerates subsequent detection steps.
Solution Approach 2:
The patent transitions from two-dimensional spatial image data to three-dimensional spatiotemporal data by incorporating the time dimension. This is achieved by generating a spatiotemporal image that combines differential images across multiple time points, enabling the system to analyze temporal patterns of movement while maintaining detection speed through efficient data representation.
2Measurement precision
If image sequence is used for moving object detection, then detection accuracy improves during occlusion and environmental fluctuations, but processing complexity increases
Solution Approach 1:
The patent applies segmentation by dividing the image sequence processing into distinct functional modules: background subtraction unit, differential image generation unit, spatiotemporal image generation unit, and moving object detection unit. Each module handles a specific aspect of the processing pipeline, reducing overall system complexity while maintaining high detection accuracy through specialized processing at each stage.
Solution Approach 2:
The patent extracts and isolates the essential moving object characteristics by removing background elements through subtraction operations. By extracting only the differential components that represent actual movement, the system reduces processing complexity while preserving detection accuracy, focusing computational resources on relevant dynamic features rather than entire image sequences.
3Ease of operation
If histogram calculation method is used, then detection is possible without object correspondence between frames, but mistaken detection increases due to lack of object characteristics and lighting changes
Solution Approach 1:
The patent introduces differential images as an intermediary representation between raw image sequences and final detection results. This intermediary step preserves object characteristics by highlighting temporal changes while filtering out static background elements and lighting variations. The differential image serves as a mediator that maintains both operational simplicity and detection accuracy by encoding movement information in a standardized format.
4Extent of automation
If geometric figure parameterization is applied to silhouette image, then movement analysis becomes possible, but parameterization precision becomes unstable due to occlusion and lighting changes
Solution Approach 1:
The patent applies dynamics by transitioning from static geometric figure parameterization to dynamic spatiotemporal pattern recognition. Instead of applying fixed geometric models to potentially distorted silhouette images, the system processes dynamic differential images that capture actual movement patterns. This dynamic approach automatically adapts to occlusion and lighting changes, maintaining both automation and precision through temporal pattern analysis rather than static geometric fitting.
Data Source
AI summary
A moving object detection device including a spatiotemporal data generation unit 120 generating time series data, which arranges data indicating a moving object along a temporal axis, based on an output from a camera 100, an inter-leg information unit 140 extracting based on the generated time series data, inter-leg information, which is information regarding a temporal change in an inter-leg area arising from movement of a moving object that has two or more legs, and a periodicity analysis unit 150 analyzing a periodicity within the extracted inter-leg information. Further, the moving object detection devices includes a moving object detection unit 160 generating, from the analyzed periodicity, movement information that includes the presence or lack thereof of a moving object.