A study on image generation with shifted perspective using two in-vehicle camera images

This paper presents a study on generation of an image using two images taken at shifted locations by in-vehicle cameras. Vehicular Ad Hoc Networks (VANETs) provide one of the most important fields of research for high interest in Intelligent Transportation Systems (ITS). As one of the ways to use VANETs, we propose an image generation method to use two images which were taken by in-vehicle cameras at different locations. The image is generated by shifted perspective projection with multiple planes. 3-D coordinates are measured and simple 3-D model made of multiple vertical planes similarly to Tour Into the Picture (TIP) is constructed by matched points. The matched points of two images are decided by Line Segment Detector (LSD) and template matching. Based on this method, an image between two images is generated. The generated image is expected to help intuitive understanding of traffic situations for drivers through VANETs. Keywords—Image processing, ITS, VANET