Occupancy detection in elevator car by fusing analysis of dual videos

Detection of occupancy in the elevator car is becoming increasingly important because the elevator control system relies the real-time data on occupancy to avoid overload, improve comfort and reduce energy consumption. This paper focuses on detection of occupancy of elevator car in the video scenes in oblique views. We propose an occupancy detection method fusing analysis of dual synchronous videos captured by two cameras installed inside and outside the elevator, respectively. One video is used to detect the absolute value of occupancy by the duration of full opening state of the elevator door. The other video is used to solve the coefficient of the occupancy by the change of foreground areas before and after the elevator door is opened. The absolute value and the coefficient jointly determines occupancy measurement. The experimental result verifies the effectiveness of our method and the correctness reaches up to 91.7%, exceeding that of the traditional method based on pixel statistics in the connected area. This method may contribute to the elevator control system after it is integrated with the video surveillance system.