Научная визуализация, 2020, том 12, номер 3, страницы 89 - 99, DOI: 10.26583/sv.12.3.08
Image monitoring and recognition processing based on neural network
Авторы: L. Min1, Y. Zhengkun 2
Changsha Vocational & Technical College, Changsha 410217, Hunan Province, China
1 ORCID: 0000-0003-2481-3789, lminlanm@yeah.net
2 ORCID: 0000-0003-0032-311X
Аннотация
With the development of economy and the abundance of material, people tend to travel. In the peak season of tourism, the scenic spots are crowded and easy to cause trample and safety problems. The traditional monitoring methods are rigid and have low recognition accuracy. This paper briefly introduced the image monitoring and recognition system and the back-propagation (BP) neural network used for identifying the trampling risk areas in the monitoring images. After that, the image monitoring and recognition system was simulated by using MATLAB software, and it was compared with the traditional entropy method and state-of-the-art CNN. The results showed that the three methods could identify the area with trampling risk in the image, but the image monitoring and recognition system designed in this study was more comprehensive and had lower false alarm rate and shorter recognition time than the traditional information entropy method and state-of-the-art CNN. In summary, the image monitoring and recognition system designed in this study can efficiently and accurately identify the trampling risk areas in the monitoring images.
Ключевые слова: image recognition, information entropy, back-propagation neural network.
