Long-Distance Small Target Recognition Based on Dynamic Frame Difference Method

As the economy continues to develop, the issue of energy wastage is becoming increasingly pronounced. In this context, an intelligent classroom lighting control system based on video content monitoring is proposed with the aim of reducing energy consumption. However, the prevailing mainstream target detection methods encounter limitations when it comes to long-distance small target recognition. To address this challenge, this paper proposes a new dynamic frame differencing method that effectively combines frame differencing with background subtraction techniques. Utilising the real-time video input from the surveillance camera, the background reference is dynamically updated using the dynamic frame difference matrix and the frame difference time series generated by the frame difference method. This approach facilitates the accurate recognition of the contours of small targets at extended distances through background subtraction. Experimental results demonstrate the efficacy of the proposed method in accurately detecting the contours of long-range small targets under varied conditions. The contour information obtained using this method can assist existing target detection techniques, such as the YOLO model, to significantly improve the recognition rate of long-range small targets. Furthermore, the effects of threshold selection and video resolution on target contour detection accuracy were analysed, providing valuable references for future research and applications.