Fisheye Image Detection Threshold Adjustment for Human Body Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Misdetections in human body detection using omnidirectional network cameras are frequent due to the fisheye lens capturing images where human bodies of similar sizes appear differently, even at close distances, making it difficult to accurately analyze images.
Innovation Solution
An image processing apparatus that obtains fisheye images, detects objects of specific sizes, and sets detection thresholds based on distance from a reference position and camera installation height to improve detection accuracy by adjusting the minimum size threshold as distance from the center increases, preventing improbable size detections.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a fixed size threshold is used for detecting human bodies in fisheye images, then the detection process is simple, but detection accuracy deteriorates because human bodies of similar sizes appear differently at different distances from the center
Solution Approach 1:
The patent applies dynamics by making the detection threshold variable rather than fixed. The threshold dynamically adjusts based on the distance from the center position in the fisheye image, allowing the detection criteria to adapt to different spatial regions where objects appear at different scales due to the fisheye lens distortion
Solution Approach 2:
The patent changes the parameter of detection threshold from a constant value to a variable value that depends on the distance from the center position. This parameter change allows the system to account for the scaling effects in fisheye images, improving detection accuracy without requiring complex additional processing
2Measurement precision
If the detection threshold is adjusted based on distance from center, then detection accuracy improves, but the detection process becomes more complex
Solution Approach 1:
The patent implements parameter changes by introducing a distance-dependent threshold parameter. The threshold is calculated based on the distance from the center position, creating a gradient of detection criteria that matches the fisheye image characteristics. This approach improves accuracy while maintaining relatively simple implementation
Solution Approach 2:
The patent applies segmentation by dividing the fisheye image into multiple regions based on distance from the center position. Each region has its own detection threshold, allowing differentiated detection strategies for different spatial zones without requiring complex global processing
Data Source
AI summary
An image processing apparatus (client apparatus) includes an obtaining unit configured to obtain a fisheye image captured by an imaging unit including a fisheye lens, a detection unit configured to detect an object having a specific size as a detection target object from the fisheye image obtained by the obtaining unit, and a setting unit configured to set a size of the detection target object to be detected by the detection unit based on a distance from a reference position in the fisheye image and a height at which the imaging unit is installed.