ANALYSIS OF FILLING OF PASSENGER TRANSPORT STOPS USING ALGORITHMS OF IMAGE PROCESSING FROM SMART CITY IP CAMERAS

Abstract

Counting the number of people at a passenger transport stop is necessary to ensure the timely delivery of vehicles of the required capacity to existing stops. In addition, it is necessary to provide for the construction of new public transport stops on the main routes of large cities in case of overloading of the existing ones. Computer systems of the "smart city" make it possible counting the number of people at the stops due to innovative technologies. For example, it is possible to use sensors that react to the presence of people, and turn on the sidewalk lights. The increasing spread of video surveillance systems within the Smart City makes it expedient to use video streams or photos from IP cameras, which are the end hosts of the information-measurement network. The analysis of the use of global (threshold) and local (adaptive) methods of binarization of images obtained from IP cameras was carried out. As a result, it is indicated that adaptive binarization can be recommended if it is necessary to process low-quality halftone images. However, in the presence of inhomogeneities of the background with low contrast, in this case the appearance of false objects is possible. At the same time, the loss of small details when using the methods of global image binarization does not affect the estimation of filling of passenger transport stops. The expediency of using Otsu's algorithm for finding the optimal threshold for image binarization in the process of passenger transportation planning is proven. The developed JavaScript application provides for the conversion of color photos from IP cameras into Grayscale images. An example of the work of the developed application when calculating the total number of pixels from photos of people at city transport stops is given.