Real-time pedestrian counting is on high demand for traffic control, marketing and management of the crowd. In this project, an automatic and robust pedestrian counting system for a shopping mall based on computer vision is proposed. It is very important to determine the recession and peak hours of the mall by visitors counting for business decision making. The proposed method utilizes a pre-installed overhead or surveillance camera on the entrance of the sidewalk or a corridor in the mall compound for security purpose. By accessing the camera, the area of interest is scanned, images are processed and the pedestrians in the regions are detected.
By using different approach of image and video processing methods, this system can be implemented in Visual Studio with the help of OpenCV image processing libraries. Several methods of image processing techniques such as background subtraction, optical flow, Harr-like feature detector and Histogram of Gradient are reviewed in this report. A pedestrian counting method based on Histogram of Gradient is presented. The principle of HOG is introduced and the detection accuracy, parameters of configuring the system and also the defectiveness of the system is being discussed. The proposed system is able to detect the positions of human accordingly to real time. The computer experiment has shown that the system is able to detect and count the human correctly..
❖ Video-Based Pedestrian Counting: Many systems rely on video cameras and computer vision algorithms to detect and track pedestrians in real-time. These cameras can be strategically placed at specific locations such as street intersections, pedestrian crossings, or public areas. The video footage is then processed using computer vision techniques to identify and count pedestrians as they pass through the monitored area.
❖ Infrared Sensor-Based Systems: Infrared sensors are often used to count pedestrians based on their heat signatures. These sensors can be mounted on walls, ceilings, or poles and are capable of detecting human presence within their range. Pedestrians walking through the sensor's field of view trigger a count, and the data is collected and analyzed by the system.
❖ Wi-Fi and Bluetooth-Based Tracking: Some systems utilize Wi-Fi or Bluetooth signals emitted by mobile devices carried by pedestrians. By capturing and analyzing these signals, the system can estimate the number of pedestrians in the area and track their movement patterns.