Najmi Hafizi, Zabawi (2016) Parking lot detection using tunable two- mode region of interest. Masters thesis, Universiti Malaysia Pahang (Contributors, UNSPECIFIED: UNSPECIFIED).
|
Pdf
Parking lot detection using tunable two- mode region of interest.pdf Download (497kB) | Preview |
Abstract
This research focuses on development of a vision-based system that able to detect the vacancy of parking spaces using image processing technique, namely through the adaptive twin ROI mode approach. In performing the parking space detection, a special computer program called ‘detector’ has been developed using the combination of OpenCV and FLTK libraries. The program which is used with web camera and notebook computer is dedicated to perform all methods that are proposed in this research. The system that has been developed is targeted to detect the parking space during daytime in outdoor environment and can work in different weather and lighting conditions such as under bright sunlight, during rainy day or in shady surroundings. The procedure in performing the parking space detection is divided into four stages namely the image acquisition, image preprocessing, ROI setting and colour detection stage. In the image acquisition stage, three ways have been identified as the method to perform the image acquisition techniques, which they are the real situation, traffic-type and offline image file technique. Next, in image preprocessing stage, the acquired image is applied with blurring effect to reduce noise and also applied with conversion function. The conversion function converts the acquired image which originally in BGR colour space into the other popular formats namely the HSV and YUV. The ROI setting stage involves assigning the position of the ROIs onto the desired parking slots and applying the range of colour that represents the parking space background colour to the ROIs which is called as ‘allowable range’ in this thesis. Finally, in the colour detection stage, the ROIs that have been properly set are performed with pixels analysis that compares each of the pixels inside the ROI with the allowable range defined in previous stage. The pixels which their colour value is inside the allowable range will be categorized as the parking space pixels and the rest will be assumed as the vehicle’s pixels which will trigger the occupation status once the threshold value is exceeded. Several methods are introduced in overcoming the issues that have arisen in this research. Some of them include the single and twin ROI mode, the AND and OR Boolean logical operators, timer function, control ROI function and finally the shadow filtering feature. All of the methods are analyzed to view their capability and efficiency in solving the arisen issues. Through the combination of the proposed methods, it can be seen that the result of the parking space detection achieved 88.9% of success rate. The rest of 11.1% failure rate is caused by the confusion of the detector with the vehicle and parking space colour and should be overcome if more checking ROIs per parking space are implemented in the future.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | Thesis (Master of Engineering (Electronics)) -- Universiti Malaysia Pahang – 2016, SV: Dr. Sunardi, Dr. Md. Rizal Bin Othman & Associate Professor Dr. Kamarul Hawari Bin Ghazali, NO CD: 10763 |
| Uncontrolled Keywords: | Parking lot; image processing technique |
| Subjects: | T Technology > TK Electrical engineering. Electronics Nuclear engineering |
| Faculty/Division: | Faculty of Electrical & Electronic Engineering |
| Depositing User: | Ms. Nurezzatul Akmal Salleh |
| Date Deposited: | 11 Jul 2017 07:08 |
| Last Modified: | 15 Dec 2021 00:02 |
| URI: | http://umpir.ump.edu.my/id/eprint/18145 |
| Download Statistic: | View Download Statistics |
Actions (login required)
 |
View Item |