ANOMALY DETECTION SMART CAMERA

Abstract

I his project report proposes a way to delect default camera events through image analysis, to ensure good image quality and the right platform for watching smveillance videos. 1 he first method removes the reduced referenced features in most legions in the surveillance image, and then detects confusing scenarios by analysing the vaiiation ol features when the image quality decreases and the viewing field changes. Recently, confusing camera detection has attracted growing interest to produce real-time camera alerts for video surveillance systems. Existing methods lor confusing camera still do not have enough power to detect a wide variety of abnormalities, and they do not have the power to improve themselves in the case of abortions by self-study. Therefore, this paper proposes moiphological analysis and in-depth reading based on an unconventional camera detection method to detect a wide variety of abnormalities. Morphological analysis is used lor easy con I using camera detection to speed up processing speed, and in depth reading is used to detect complex camera distractions to improve accuracy. 1 est results show that the accuracy of the proposed acquisition method gains than 95%. more The system starts with the previous process of video capture with limited video, inverting and sliding. Sorting, sorting, resizing and extracting features also done in this process. Next, a network feed process is requested to generate output matrix. Based on the output matrix, the known character can be determined. This program is designed to customize the network to each user. Recommendations for future development and conclusions are also included in the report.