Omnidirectional Camera Passage Detection Using Foot Portion Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional monitoring camera systems using omnidirectional cameras with fish-eye lenses face detection errors when objects pass near passage detection lines due to the size and distortion issues, particularly in capturing narrow spaces.
Innovation Solution
An image capturing apparatus is designed with an object detection unit, a setting unit to establish a vertical reference position, and a passage detection unit that determines object passage by analyzing the relationship between the object's foot portion and a detection line set on the captured image, reducing detection errors by ensuring both elements are on the same plane.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If an omnidirectional camera with fish-eye lens is used to monitor a wide space, then the monitoring coverage is improved, but detection errors occur when objects pass near the passage detection line due to distortion and size effects
Solution Approach 1:
The patent introduces a vertical dimension by setting a determination position at the foot portion of the object (lowest point in vertical direction) rather than using horizontal crossing detection. This dimensional shift allows accurate passage detection despite the distortion inherent in omnidirectional imaging, as the foot portion position can be reliably determined even when the object appears distorted in the wide-angle view.
Solution Approach 2:
The patent replaces traditional mechanical passage detection methods (detecting when an object crosses a line) with an image processing approach that identifies the foot portion of the object based on its position in the captured image. This substitution allows for more accurate detection in distorted omnidirectional views by using computational methods rather than simple line-crossing logic.
2Extent of automation
If a passage detection line is set on the captured image, then passage detection is enabled, but detection errors occur when objects partially cross the line due to their size
Solution Approach 1:
The patent applies local quality by focusing detection on a specific local feature of the object (the foot portion) rather than requiring the entire object to cross the detection line. This localized approach ensures that even small objects or objects at various distances are detected accurately, as long as their foot portion position can be determined in the image.
Solution Approach 2:
The patent introduces the concept of a determination position (foot portion) as an intermediary between the object and the detection line. Instead of directly detecting line crossing, the system first identifies the foot portion position and then determines passage based on the relationship between this intermediary position and the detection line, improving reliability.
Data Source
AI summary
An image capturing apparatus comprises an object detection unit configured to detect a region of an object from a captured image captured by an image capturing unit via an omnidirectional imaging lens; a setting unit configured to set a position in the captured image corresponding to a direction perpendicular to a horizontal plane; a decision unit configured to decide, based on a relationship between the set position and a position of the object region in the captured image, a determination position used to determine passage of the object out of the region of the object; and a passage detection unit configured to detect, based on the determination position decided by the decision unit and a detection line set on the captured image, passage of the object across the detection line.