Research And Application Of Face Recognition Based On Video Surveillance In Inspection Management Of Hydropower Station

Hydropower plays the vital role of a power supply,peak shaving,and frequency modulation.The equipment safety of hydropower station is directly related to the normal operation of the power grid.It is a crucial project.At the hydropower station,the operation and maintenance team will be patrolled for on-site equipment.The investigation found that the current inspection site adopts the sign-in mode to conduct inspections in place,which is not suitable for inquiries,missed inspections,private shifts,etc.To effective auxiliary supervision.Given the above problems,this paper proposes the face detection and identification of surveillance video and the oversight of the inspection process of the inspection personnel from the perspective of economic efficiency,combined with the existing surveillance cameras and face recognition theory.Firstly,the characteristics of a hydropower station monitoring video are analyzed.For the noise generated by video images in the scene environment,several classical image denoising methods are used for denoising,and the results are compared.The wavelet denoising method is selected for noise filtering.Besides,the image contrast problem is analyzed according to the scene lighting environment.The multi-scale Retinex algorithm is used to enhance the image contrast,which provides a guarantee for subsequent face detection and recognition.In the face detection process,the Adaboost algorithm is used to train the cascade classifier,and experiments are carried out on the laboratory surveillance camera.The false detection and miss detection of the detection algorithm are found.The following solutions are proposed: the existence of the target is confirmed by continuous detection,the false detection phenomenon is reduced,and the influence of motion blur on the recognition result is also prevented.Then,the resolution of the image is proposed by using the definition evaluation function for the multi-frame detection result.Evaluation,push the background for face recognition with optimal results,improve the system recognition rate,and reduce the computational time complexity of the system.For the face recognition process,the PCA algorithm was introduced and analyzed,and the 2DPCA algorithm improved for PCA algorithm was introduced.The experiment was carried out in the ORL face database and its database.Finally,the face recognition system is designed according to the requirements.The system is based on MATLAB.The system interface is simple,and the orientation is strong.The patrol personnel is determined according to face recognition,and the inspection time isextracted.Combined with the surveillance camera address,the inspection location is determined.The system requirements are realized,and the output report form is adopted,and the personnel inspection time sheet is output.