Research on Pedestrian Intrusion Detection Method in Coal Mine Based on Deep Learning

Due to the complex background environment in coal mines, the timeliness and accuracy of pedestrian intrusion detection are low. In order to improve the detection accuracy and efficiency of pedestrian detection in complex coal mines, a deep learning-based pedestrian intrusion detection method in coal mines was studied. Build a pedestrian intrusion detection model in coal mine, the grayscale, denoising and illumination equalization processing is carried out for the surveillance video images of pedestrians in the coal mine. The image is preprocessed by nonlinear transformation method, gradient descriptor is obtained by gradient calculation method, HOG feature is obtained, and texture feature is obtained by LBP operator, and the features are used as input to construct a detection model using the restricted Boltzmann machine in deep learning to realize pedestrian intrusion detection in coal mines. The experimental results show that under the application of the research method, the average accuracy rate is higher, reaching more than 90%, and the FPS value is greater, reaching more than 40fps, indicating that the research method has higher detection accuracy and faster detection speed.