Vision-based customer counting for shopping lots
Market research is vital for businesses to prosper in this modern world, and customer traffic data forms a very important part of the market research studies in order for business owners to gauge the attractiveness of the products available at their business premises and hence allow them to strategi...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2014
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/id/eprint/47991/25/LeeFangSoongMFKE2014.pdf http://eprints.utm.my/id/eprint/47991/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Market research is vital for businesses to prosper in this modern world, and customer traffic data forms a very important part of the market research studies in order for business owners to gauge the attractiveness of the products available at their business premises and hence allow them to strategize their business or marketing plan accordingly. However, the problem arises when customer traffic data is collected manually, which is tedious, time-consuming and prone to error. The purpose of this project is to develop a simple, user-friendly software based on MATLAB to count the number of people at a premise. The input to the software is a pre-recorded video sequence captured by an overhead camera hung from the ceiling. This video sequence involves top view scenes of people walking at a designated area in the business premises within the sight of an overhead camera. In order to perform the task of counting the number of people in a video, multiple digital image processing techniques are implemented, starting with frame grabber to extract still images from a video sequence, followed by image pre-processing steps including RGB image conversion to grayscale, background subtraction, threshold-based object segmentation and morphology operations. The last step involves tracing the boundary of the region of interest and then label and count it based on blob measurement. The project integrates these steps into a flow within a Graphical User Interface (GUI), and the final implementation is a working GUI with the capability to accept user interaction. The completed software is able to count the number of people, with accuracy at about 81% depending on the image conditions. |
|---|