Surveillance Object Detection via Clustering and Edge Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image surveillance systems struggle to detect objects left behind or taken away in complex environments with varying lighting conditions and multiple moving objects, often resulting in false alarms and reduced performance.
Innovation Solution
An image surveillance system that includes a foreground detecting unit, a still region detecting unit, and an object detecting unit, which utilize pixel information differences, clustering, and edge analysis to differentiate between true and falsely detected still regions, and determine if an object is left behind or taken away based on edge intensity changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If related art methods are used to detect objects left behind or taken away, then detection can be performed in simple environments, but performance is remarkably reduced in complex environments with multiple moving objects and lighting changes
Solution Approach 1:
The patent segments the detection process into multiple independent stages: foreground detection, still region detection with clustering, false-detection determination using multiple criteria, and object detection. Each stage processes specific aspects of the image data independently, allowing the system to handle complex environments by breaking down the overall detection task into manageable sub-tasks that can be optimized individually.
Solution Approach 2:
The patent applies different detection criteria and thresholds to different regions and characteristics of the image data. The false-detection determining unit evaluates multiple specific characteristics (area ratio, variance in area, coordinate variance, sustained time period, size, motion, darkness change, average pixel value relationship) with different weights and thresholds appropriate to each characteristic, allowing optimal detection performance for each local aspect of the surveillance data.
2Device complexity
If simple detection methods are used, then the system is easier to implement, but it generates many false alarms in complex environments with lighting changes and multiple moving objects
Solution Approach 1:
The patent performs preliminary detection and filtering actions before final object detection. The still region detecting unit first identifies candidate still regions, then the false-detection determining unit preliminarily filters these candidates by evaluating multiple criteria (area ratio, variance, time period, etc.) to eliminate false detections before the final object detection stage, reducing false alarms early in the process.
Solution Approach 2:
The system incorporates feedback mechanisms where the false-detection determining unit continuously monitors and evaluates candidate still regions against multiple criteria, and the object detecting unit uses feedback information about true and false detections to adjust and refine its detection decisions, thereby reducing false alarms through iterative refinement.
Data Source
AI summary
An image surveillance system and a method of detecting whether an object is left behind or taken away are provided. The image surveillance system includes: a foreground detecting unit which detects a foreground region based on a pixel information difference between a background image and a current input image; a still region detecting unit which detects a candidate still region by clustering foreground pixels of the foreground region, and determines whether the candidate still region is a falsely detected still region or a true still region; and an object detecting unit which determines whether an object is left behind or taken away, based on edge information about the true still region.