Head-Counter Device Noise Area Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing head-counter devices face challenges in accurately counting individuals in crowded and open spaces due to structural complexity and inefficiencies, particularly in outdoor environments where noise areas like sky, vegetation, and buildings interfere with image processing.
Innovation Solution
A head-counter device equipped with a digital camera and image processing module that automatically identifies and removes noise areas such as sky, vegetation, and building facades from images, allowing for precise counting and profiling of individuals through cropping and resizing, utilizing algorithms like R-FCN networks for efficient processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the whole image area is processed for person counting, then the counting coverage is comprehensive, but the computational complexity and processing time increase significantly
Solution Approach 1:
The image is divided into multiple sub-regions based on depth information from the depth map. The processing area is segmented to exclude sky, vegetation, and building facades, focusing computational resources only on ground-level areas where persons are likely to be present. This segmentation reduces the number of pixels requiring complex person detection algorithms while maintaining comprehensive counting coverage in relevant areas.
Solution Approach 2:
Noise areas (sky, vegetation, building facades) are extracted and removed from the processing area using depth-based segmentation. By taking out these irrelevant regions, the system reduces computational burden and processing time while preserving the ability to accurately count persons in the remaining ground-level areas.
2Measurement precision
If the whole image area is processed for person counting, then the counting coverage is comprehensive, but the processing time increases significantly
Solution Approach 1:
The image processing is segmented into depth-based regions, allowing the system to process only ground-level areas where persons are present. This segmentation eliminates the need to analyze sky, vegetation, and building facades, significantly reducing processing time while maintaining accurate person counting in relevant areas.
Solution Approach 2:
A depth map is generated as a preliminary step before person detection. This depth information is used to pre-segment the image and identify noise areas that should be excluded from processing. By performing this preliminary segmentation, the system avoids wasting processing time on irrelevant regions during the main person counting operation.
3Ease of manufacture
If noise areas like sky, vegetation, and building facades are included in processing, then the image processing is simple, but the person counting accuracy decreases in crowded and limited areas
Solution Approach 1:
The processing approach dynamically adapts based on depth information. The system automatically adjusts the processing area by excluding noise regions identified through depth-based segmentation. This dynamic adaptation maintains processing simplicity through automated depth-map generation while significantly improving counting accuracy by focusing computational resources on ground-level areas where persons are present.
Data Source
Figure 1
Figure 2
AI summary
A head-counter device (100) comprising a digital camera (1) adapted to provide a first digital image (IM1) representative of a counting zone (STR) of persons, the first image defining a first horizontal dimension (N) and a first vertical dimension (M), and a cropping module (7) configured for: analyzing the first image (IM1) and identifying a noise area (PCR) according to at least one of the following features: pixel light intensity, pixel color and/or presence of predefined patterns, cropping the noise area (PCR) from the first image (IM1) to obtain a second image (IM2) without the noise area, the noise area (PCR) being a peripheral portion of the first image having said first horizontal dimension and having a second vertical dimension (M-DSK+SM) shorter than the first vertical dimension.