Ground plane velocity estimation embedding rectification on a particle filter multi-target tracking

Abstract

This paper presents an integrated solution for vehicle's velocity estimation and vehicle counting. The proposed restores the scene geometric properties, building a ground plane rectified image. Moreover, multiple vehicles tracking is performed embedding the concept of region covariance descriptors in a particle filter framework. The results show the effectiveness of the approach here proposed in very clutter scenes.

Parallel Implementation of a Video-based Vehicle Speed Measurement System for Municipal Roadways

“…Similarly, in [16] a frame differencing and blob tracking method is presented in which the vehicle's speed is obtained by estimating each blob's displacement using static parameters. In other approaches, authors of [17][18][19] used a median filter for moving vehicle detection and calculated the real position of the vehicle in video frames to measure speed. A Gaussian distribution for detecting moving vehicles is presented in [20] which uses blob detection and tracking for speed calculation.…”

Vehicle speed estimation by license plate detection and tracking

“…There is an extensive literature on vehicle speed estimation [6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24]. Most approaches are based on specialized speed sensors [6], or are dedicated to specific contexts, such as speed estimation from vehicle's headlight in night scenes [7,8], speed estimation using the signal from mobile telecommunication network [9], or speed estimation from a car-mounted camera to avoid collisions [10].…”

“…In 2009, Palaio et al [16] proposed to track vehicles using particle filtering. The vehicles are selected by a foreground/background segmentation process and are then represented by a descriptor based on window location, window color components, horizontal and vertical derivatives, and the Laplacian of the grey image.…”

OCULUS surveillance system: Fuzzy on-line speed analysis from 2D images

“…Once these images have been generated, the displacements of the vehicles are analysed to estimate their speeds in a third phase. Other similar methods in which the boundary lines of a highway are automatically detected, are discussed in Beymer et al (1997), Palaio et al (2009). Maduro et al described in Maduro et al (2008) how the previously cited methods lose their effectiveness when there are occlusions between objects, long distances covered by vehicles in few frames or low image resolution, among other reasons.…”

“…Many of these methods are computationally expensive and the use of several cameras is an expensive solution. Other authors have proposed methods to overcome the same problem by using a single camera (Beymer, McLauchlan, Coifman, & Malik, 1997;Cathey & Dailey, 2005;Maduro, Batista, Peixoto, & Batista, 2008;Palaio, Maduro, Batista, & Batista, 2009). In this kind of methods, a calibration process is normally carried out in which parameters such as height, zoom level or angle camera must be previously known.…”