Seat-belt Detection Using Image Processing in MATLAB

Abstract

In this study, a method is suggested to detect seat belts in a monitoring image of a moving vehicle. The method relies on several image processing techniques such as edge detection, closing operation, and Hough transform, to precisely identify the seat belts. The proposed method for detecting seat belts is more dependable and sturdier compared to the methods that previously depended on driver's visual monitoring or basic belt detection systems. The initial step of the algorithm involves detecting the lines present in the input image, achieved through the utilization of the Hough transform. Then, the detected lines are filtered based on their orientation and location, allowing only the lines that are likely to represent seat belts to be selected. The closing operation is then used to connect the selected lines and form a complete seat belt. Finally, the algorithm verifies the seat belt by checking the continuity and width of the detected line. The proposed method was evaluated on a dataset of monitoring images containing various driving scenarios, and the results show high accuracy in seat belt detection. The algorithm was able to detect seat belts with a high true positive rate demonstrating its effectiveness in real-world scenarios. Overall, the proposed method provides a reliable and accurate approach to seat belt detection, which can be integrated into existing monitoring systems in vehicles to improve safety. This approach underscores the potential of technology to improve road safety and mitigate the likelihood of accidents, as demonstrated through the application of image processing techniques.