Fisheye Lens Human Body Detection via Priority Area
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Wide-angle image pickup systems, such as those using fisheye lenses, face challenges in efficiently detecting human bodies due to image distortion, which increases processing time and can lead to inaccurate detection if the camera installation angle is not correctly set, and existing methods either require raster-scanning the entire image or fail to extract features from areas without abnormalities.
Innovation Solution
An image processing apparatus that determines the camera installation direction and prioritizes human body detection in areas where the installation direction indicates a higher likelihood of human presence, using distortion correction methods like double panoramic or panoramic conversion based on the installation direction, allowing for efficient detection without scanning the entire image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the entire image is scanned to detect human bodies in wide-angle images, then detection completeness is improved, but detection time increases significantly
Solution Approach 1:
The patent divides the wide-angle image into multiple regions based on distortion characteristics, with the central region having lower distortion and peripheral regions having higher distortion. By segmenting the image and applying different detection strategies to different regions, the system achieves both comprehensive detection and reduced processing time.
Solution Approach 2:
The patent applies different weight coefficients to different regions of the image based on their distortion characteristics. The central region is assigned a weight coefficient of 2, while peripheral regions are assigned a weight coefficient of 1. This local quality adjustment optimizes detection efficiency by focusing more computational resources on regions where human bodies are more likely to be detected accurately.
2Measurement precision
If distortion correction is applied to the entire image, then image accuracy is improved, but processing complexity increases
Solution Approach 1:
The patent segments the image into central and peripheral regions and applies distortion correction selectively. The central region undergoes standard distortion correction, while peripheral regions use simplified correction methods or are processed with different parameters, thereby reducing overall processing complexity while maintaining image accuracy where it matters most.
Solution Approach 2:
Different distortion correction parameters and methods are applied to different regions of the image. The central region receives full correction with higher priority, while peripheral regions receive adjusted correction with lower priority, optimizing the balance between image accuracy and processing complexity.
3Measurement precision
If weight coefficients are applied to all image areas for human body detection, then detection accuracy in peripheral areas is improved, but overall processing time increases
Solution Approach 1:
The patent applies different weight coefficients to different regions: weight coefficient 2 for the central region and weight coefficient 1 for peripheral regions. This local quality differentiation improves detection accuracy in peripheral areas while avoiding the time penalty of uniformly high-weight processing across the entire image.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
It is an object of the invention to efficiently detect a recognition target on an image photographed through a fisheye lens. In accordance with an installation direction of a camera, an area where a relatively large number of human bodies are detected in an area of the photographed image is set as a priority area. A detecting process of the human body is preferentially started from the set priority area.