Detection and measurement of obstacles on a track using color segmentation with background subtraction and morphological operation

Abstract

The process of detecting objects in a path of the image frame is a complex process. Generally, the identification process to find out the condition of a track in the form of objects and the size of objects is done manually by humans, namely using measuring instruments such as rulers, etc. It requires long processing time, and is less efficient, besides that it cannot be done remotely in the form of monitoring on the computer from the camera in terms of measurement. This study aims to create a system of detection and measurement of obstacle on a track using color segmentation method with background subtraction and morphological operation as a result of decision support. This detection and measurement system consists of webcam image acquisition, video extraction, converting frames of RGB to HSV images, color segmentation with background subtraction, gray scaling and thresholding, morphological analysis to improve the results of segmentation, detection and measurement displayed in the form of a Graphical User Interface (GUI). The performance of the system was tested on two background color references in the form of black, namely the asphalt track and white, namely the concrete cement track. The result of testing with the black background, produces an accuracy value of 96%, and with a white background of 100%. The system results in measuring obstacle have an average length of accuracy of 96.16% and a height of 85% on a black background in the form of an asphalt track, and on a white background in the form of cement concrete track results in a length measurement accuracy of 95.99% and a height of 90.88% with the camera moving or fixed.