Distracted driving is one of the main causes vehicle collisions in United States. Passively monitoring a driver's activities constitutes basis an automobile safety system that can potentially reduce number accidents by estimating focus attention. This paper proposes inexpensive vision-based to accurately detect Eyes Off Road (EOR). The has three components: 1) robust facial feature tracking; 2) head pose and gaze estimation; 3) 3-D geometric reasoning EOR. From video stream camera installed...

In this paper, a LIDAR-based road and road-edge detection method is proposed to identify regions road-edges, which an essential component of autonomous vehicles. LIDAR range data decomposed into signals in elevation projected on the ground plane. First, elevation-based are processed by filtering techniques candidate region, pattern recognition determine whether region segment. Then, line representation plane identified compared simple model top-down view segment with its road-edges. The...

We describe a real-time pedestrian detection system intended for use in automotive applications. Our demonstrates superior performance when compared to many state-of-the-art detectors and is able run at speed of 14 fps on an Intel Core i7 computer applied 640×480 images. approach uses analysis geometric constraints efficiently search feature pyramids increases accuracy by using multiresolution representation model detect small pixel-sized pedestrians normally missed single approach. have...

Multi-attribute decision-making problems under the trapezoidal fuzzy neutrosophic numbers environment are complex, particularly when attribute value data incomplete, and weight is completely unknown. As a solution, this study proposes method based on information entropy grey theory. First, regarding problem of incompleteness in decision matrix, defining missing using proposed, which simpler more reasonable than traditional complements. Regarding unknown weights, definition new proposed used...

Bicycles that share the road with intelligent vehicles present particular challenges for automated perception systems. Bicycle detection is important because bicycles and can move at comparable speeds in urban environments. From a computer vision standpoint, bicycle challenging as bicycle's appearance change dramatically between viewpoints person riding on non-rigid object. In this paper, we vision-based framework to detect track takes into account these issues. A mixture model of multiple...

This paper presents a vision-based framework for intelligent vehicles to detect and track people riding bicycles in urban traffic environments. To deal with dramatic appearance changes of bicycle according different viewpoints as well nonrigid nature human appearance, method is proposed which employs complementary detection tracking algorithms. In the phase, we use multiple view-based detectors: frontal, rear, right/left side view. For each view detector, linear Support Vector Machine (SVM)...

Raindrops on vehicles' windshields can degrade the performance of in-vehicle vision systems. In this paper, we present a novel approach that detects and removes raindrops in captured image when using single camera. When driving light or moderate rainy conditions, appear as small circlets windshield each frame. Therefore, by analyzing color, texture shape characteristics images, first identify possible raindrop candidates regions interest (ROI), which are locally salient droplets saliency...

This paper presents a monocular vision based 3D bicycle tracking framework for intelligent vehicles on detection method exploiting deformable part model and using an Interacting Multiple Model (IMM) algorithm. Bicycle is important because bicycles share the road with can move at comparable speeds in urban environments. From computer standpoint, challenging as bicycle's appearance change dramatically between viewpoints person riding non-rigid object. To this end, we present...

Detecting objects in shadows is a challenging task computer vision. For example, clear path detection application, strong on the road confound of boundary between and obstacles, making algorithms less robust. Shadow removal, relies classification edges as shadow or non-shadow edges. We present an algorithm to detect edges, which enables us remove shadows. By analyzing patch-based characteristics (e.g., object edges), proposed detector can discriminate from other images by learning...

For the multiple attribute group decision making (MAGDM) problem, in which weights are unknown and value of alternatives is form a trapezoidal fuzzy neutrosophic number, this paper proposes two methods: one based on number hybrid averaging (TrFNNHA) operator, other technique for order performance by similarity to ideal solution (TOPSIS) method. First, obtained using truth favorite relative expected value, distance measure defined cosine measure. Next, proposed ordered weighted arithmetic...

In recent years, typhoon disasters have occurred frequently and the economic losses caused by them received increasing attention. This study focuses on evaluation of based interval neutrosophic set theory. An (INS) is a subclass (NS). However, existing exponential operations their aggregation methods are primarily for intuitionistic fuzzy set. So, this paper mainly focus research operational laws numbers (INNs) in which bases positive real exponents numbers. Several properties law discussed...

Abstract This paper proposes a group decision making method based on entropy of neutrosophic linguistic sets and generalized single valued operators. is applied to solve the multiple attribute problems under liguistic environment, in which weights are completely unknown. First, obtained by using sets. Then three operators introduced, including weighted averaging (GSVNLWA) operator, ordered (GSVNLOWA) operator hybrid (GSVNLHA) GSVNLWA GSVNLHA used aggregate information. Furthermore,...

In this paper, we describe a prior-based vanishing point estimation method through global perspective structure matching (GPSM). contrast to the traditional approaches which require an undistorted image with straight roads for estimation, our first infers candidates of input from database pre-labeled points. An image-based retrieval is used identify best candidate images in by image's structure. The initial calculated points candidates. Probabilistic refinement (PR) then optimize estimate....

To be deployed in the real world, a self-driving car must capable of responding to exceptional road conditions, such as temporary work zones, because events can change previously known traffic rules and geometry. develop capability, we implemented computer vision system that recognizes bounds highway workzones by detecting regulatory warning workzone signs. Because it is not practical expect perfect performance sign recognition, also developed confidence-propagation method handle potential...

This paper presents an automated monocular-camera-based computer vision system for autonomous self-backing-up a vehicle towards trailer, by continuously estimating the 3D trailer coupler position and feeding it to control system, until alignment of tow hitch with trailers coupler. is made possible through our proposed distance-driven Multiplexer-CNN method, which selects most suitable CNN using estimated coupler-to-vehicle distance. The input multiplexer group detector, trackers, localizer....

With the goal of avoiding obstacles on road using only a single camera during autonomous driving, we propose an example-based clear path detection method that also considers additional perspective cue from estimated vanishing point. First, because benefit point provides knowledge for detection, apply to estimate initial candidates. Then, instead building pre-trained model over limited training set, global image matching get approximate idea candidate regions, and use Gaussian Mixture Model...

Trapezoidal fuzzy neutrosophic decision making plays an important role in decision-making processes with uncertain, indeterminate, and inconsistent information. In this paper, we propose a new multi-attribute method based on trial evaluation laboratory (DEMATEL), distance, linear assignment (LAM), express values as the trapezoidal numbers (TrFNNs). First, attribute weights are obtained using DEMATEL distance of TrFNNs graded mean integration representation is defined. Then, alternatives...

This paper proposes a novel system for automatically detecting children from color monocular back-up camera, as part of warning device in passenger vehicles. We presented the use an attentional mechansim that focuses compute-intensive bounding-box classifiers on subset all possible solutions to enable real-time performance 248ms per frame with negligible reduction performance. The mechanism called Attention Children which consists window generation and verification cascade based...