Analysis of image transfer mechanisms in a RESTful API client-server architecture and its application to lane detection

This study investigates the effect of image rescaling and format changing hosted in a RESTful API architecture considering transfer time as a critical factor to apply lane detection on a external server. We employ Probabilistic Hough Transform to perform the detection because it is invariant to pixel position and robust to noise. Before being trasmitted to the external server, images are rescaled and their formats are changed, this way lane detection on the images can be performed faster than working with the original images. From the image formats used in the experiments, the JPG or JPEG yields the best results. When performing the rescaling and format change, transmission and processing time is halved.