Object Detection Using Image Discontinuousness for Low-Resolution Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object detection methods, such as Histograms of Oriented Gradients (HOG), require high image resolution and are computationally intensive, making them inefficient for real-time applications and low-quality images, often failing to detect complete objects due to noise and blurring.
Innovation Solution
An enhanced object detection method utilizing image discontinuousness, which computes edge strength and builds an optimal edge path to identify objects by marking areas with high discontinuousness values, reducing the need for high image quality and computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If HOG is used for object detection, then detection accuracy can be improved, but computational complexity increases and real-time performance deteriorates
Solution Approach 1:
The patent segments the object detection process into two stages: first using a simple edge-based discontinuousness method to identify candidate regions, then applying more sophisticated analysis only to these marked areas. This segmentation reduces overall computational complexity while maintaining detection accuracy by avoiding exhaustive analysis of the entire image.
Solution Approach 2:
The patent applies partial action by using a simplified detection approach (edge discontinuousness) for the initial screening phase, and only applies the more computationally intensive HOG analysis to the marked candidate regions. This partial application of complex methods reduces total computation while preserving accuracy where needed.
2Measurement precision
If HOG is used for object detection, then detection accuracy can be improved, but detection speed decreases
Solution Approach 1:
The detection process is segmented into a fast preliminary stage using edge discontinuousness marking and a slower detailed analysis stage using HOG only on marked regions. This segmentation improves detection speed by avoiding full HOG computation on entire images while maintaining accuracy through focused analysis of candidate areas.
Solution Approach 2:
The patent performs preliminary action by first marking candidate regions using the computationally simple edge discontinuousness method before applying HOG analysis. This preliminary marking step filters out non-candidate areas, enabling faster overall detection while preserving accuracy in the final classification stage.
3Measurement precision
If high image resolution is used for HOG detection, then detection accuracy improves, but the method fails under low image quality conditions
Solution Approach 1:
The patent changes the detection parameter from gradient orientation histograms (HOG) to edge discontinuousness measurement. This parameter change enables effective detection under varying image quality conditions because edge discontinuousness is more robust to noise and resolution variations compared to HOG features, thereby improving adaptability while maintaining accuracy.
4Measurement precision
If HOG is used for object detection, then complete object detection can be achieved, but computation time increases significantly
Solution Approach 1:
The patent performs preliminary marking of candidate regions using edge discontinuousness before applying HOG analysis. This preliminary action reduces computation time by limiting HOG processing to only the marked candidate areas rather than processing the entire image, while still achieving complete object detection through the two-stage approach.
Solution Approach 2:
The detection process is segmented into a fast candidate identification phase and a slower verification phase. This segmentation reduces computation time by applying the computationally intensive HOG method only to marked candidate regions identified by the simpler edge discontinuousness method, while maintaining detection completeness.
Data Source
AI summary
An enhanced object detection method uses image discontinuousness for enhancing performance of identifying objects of a specific class in an image data. The method includes retrieving an image data; computing an image discontinuousness value between a first area and other areas surrounding of the first area, which is with different sizes and in different positions within the image data, and marking areas with an image discontinuousness value larger than a threshold; and identifying the objects of the specific class within the sliding window and outputting detection result.