Moving Object Detection via Histogram Smoothing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing moving object detection technologies, such as the dynamic contour model (SNAKES), struggle to accurately identify individual objects among multiple neighboring objects and are influenced by local noise in edge pixel histograms, particularly when objects like humans raise their hands, causing misidentification of the object center.
Innovation Solution
A moving object detection apparatus and method that utilizes synchronized cameras to generate distance and motion information, creating an object distance image and determining an object image area by smoothing pixel histograms in the vertical direction, allowing precise detection of moving objects by identifying the most symmetric point in the smoothed histogram, even when local maxima or minima are present.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If histogram smoothing is not applied, then local noise in edge pixel histograms affects detection accuracy, but applying smoothing increases computational complexity
Solution Approach 1:
The patent applies histogram smoothing as a preliminary processing step before object center detection. By pre-smoothing the histogram of edge pixel counts in the vertical direction, the method eliminates local noise peaks that would otherwise interfere with accurate center detection, while keeping the overall computational burden manageable through efficient smoothing algorithms
2Device complexity
If conventional contour detection is used on multiple neighboring objects, then detection process is simple, but multiple objects are recognized as a single body
Solution Approach 1:
The patent segments multiple neighboring objects by analyzing the histogram of edge pixel counts in the vertical direction. By identifying peaks and valleys in this histogram, the method determines horizontal division lines that separate adjacent objects, allowing each object to be individually detected and processed even when they are physically touching or overlapping in the image
Solution Approach 2:
The patent transitions from two-dimensional spatial contour analysis to one-dimensional histogram analysis in the vertical direction. By projecting edge pixel counts onto the vertical axis and analyzing the resulting histogram, the method extracts object separation information that is not apparent in the original 2D image space, enabling accurate individual object detection
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables precise detection of moving objects by eliminating local noise and accurately determining the center of moving objects, even when they are close or in the same distance, improving the accuracy of object identification and reducing computational load.
Implementation Method 1
distance information generating means that generates distance information regarding distance to the image acquisition objects on a basis of parallax of the cameras
Implementation Method 2
motion information generating means that generates motion information regarding motion of the moving objects on a basis of differences between video images input in time-series by one of the cameras
Data Source
AI summary
The present invention detects a moving object by generating the distance information of the moving object, detecting the object motion, determining the object distance, detecting the object image area and the object contour from the video image that includes the object image and contour, and provides a moving object detection apparatus to carry out such detection as well as detecting a contour of the specific moving object by detecting the center of the moving object in high precision.