Object Crossing Detection Using Relative Size Ratios
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic surveillance systems using motion video analysis for detecting objects crossing predetermined lines often generate false positives due to the inability to distinguish object sizes accurately, leading to cumbersome filter settings and adjustments.
Innovation Solution
A method involving the use of two virtual lines defined by coordinates in the scene, where the distance between them varies with depth, allowing for the calculation of an object's size relative to its distance from the camera, thereby generating an event signal only when the object's size is within a predetermined range, reducing false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If minimum and maximum size filters are implemented to eliminate false positives, then detection reliability is improved, but system complexity and ease of operation deteriorate due to cumbersome filter setup and adjustment
Solution Approach 1:
The patent changes the parameter used for object filtering from absolute size thresholds to relative size ratios. Instead of setting fixed minimum and maximum size filters that require manual adjustment, the system calculates the ratio between an object's size and the distance between two reference lines. This relative parameter automatically adapts to different scenes and camera positions, eliminating the need for manual filter setup while maintaining high detection reliability.
2Reliability
If fixed size filters are used to distinguish objects of interest, then false positives are reduced, but adaptability to different scenes and distances deteriorates
Solution Approach 1:
The system transitions from using fixed size parameters to relative size parameters. By calculating the ratio between object size and the distance between two depth-varying reference lines, the system creates a scale-invariant measurement that automatically adapts to different scenes, camera positions, and object distances, while maintaining consistent false positive reduction across all conditions.
Solution Approach 2:
The patent introduces a depth dimension by using two reference lines at different distances from the camera. The distance between these lines varies with depth, creating a three-dimensional reference framework. This allows the system to calculate object sizes in a way that accounts for perspective and distance, making the detection adaptable to various scenes without requiring manual reconfiguration.
Data Source
AI summary
A method for detecting an object crossing event at a predetermined first line in a scene captured by a motion video camera is disclosed. The method comprises determining from images of the scene captured by the motion video camera if an object image crosses the predetermined first line, calculating a size value relating to the size of the object image crossing the predetermined first line, setting a line distance value to a value calculated from the distance between a contact point of the object with the predetermined first line and a nearby point on a predetermined second line, and generating an object crossing event signal if a relation between the calculated size value and the line distance is within a predetermined range.