Road intersection detection and classification using hierarchical SVM classifier

Abstract

In this paper, a hierarchical multi-classification approach using support vector machines (SVM) has been proposed for road intersection detection and classification. Our method has two main steps. The first involves the road detection. For this purpose, an edge-based approach has been developed using the bird's eye view image which is mapped from the perspective view of the road scene. Then, the concept of vertical spoke has been introduced for road boundary form extraction. The second step deals with the problem of road intersection detection and classification. It consists on building a hierarchical SVM classifier of the extracted road forms using the unbalanced decision tree architecture. Many measures are incorporated for good evaluation of the proposed solution. The obtained results are compared to those of Choi et al. (2007).

A Survey on Outdoor Navigation Applications for People With Visual Impairments

“…With the rise of AI for automated navigation, machine learning can now be used to digitise physical street features into map information. To date, three machine learning approaches have been used by researchers in the AI domain to address the problem of intersection detection: classification [19], [20], [21], [22], road detection and intersection detection [16], [23], [24], and object detection [25].…”

A Systematic Review of Urban Navigation Systems for Visually Impaired People

“…An image classification problem: researchers have treated the problem as three levels of classification: a binary problem of existence of an interface, a multi-class intersection type problem, and a road detection problem. This latter approach is about detecting a road in an image, and then determining intersections as part of road detection [ 87 , 91 ]. Looking at each in turn, for binary classification: Kumar et al [ 88 ] determined the existence of an intersection in a video or not—the network consists of Convolutional Neural Network (CNN), bi-Long short-term memory (LSTM), and Siamese-CNN.…”

“…Finally, they integrated the two outputs to define seven classes of junctions. The third approach, identify road before classification , both Rebai et al [ 91 ] and Tümen and Ergen [ 87 ] depend on different edge-based approaches to detect the road prior to the classification step. For a classification step, Rebai et al [ 91 ] used a hierarchical support vector machine (SVM), while Tümen and Ergen [ 87 ] applied a CNN network.…”

Low Complexity Lane Detection Methods for Light Photometry System

“…There are many different variations of these approaches as well as their combinations. As stated in the article [10], one of the most robust methods is still the Hough transform and basic image processing. The road boundaries detection algorithm is based on edge detection.…”