Science
Computer Vision and Image Understanding 117 (2013) 29–41
Contents lists available at SciVerse ScienceDirect
Computer Vision and Image Understanding journal homepage: www.elsevier.com/locate/cviu
Adaptive Census Transform: A novel hardware-oriented stereovision algorithm q
Stefania Perri, Pasquale Corsonello ⇑, Giuseppe Cocorullo
Department of Electronics Computer Science and Systems, University of Calabria, Rende, Italy
a r t i c l e
i n f o
Article history:
Received 21 March 2012
Accepted 6 October 2012
Available online 17 October 2012
Keywords:
Stereovision algorithm
VLSI
Disparity Map Calculation
a b s t r a c t
This paper presents a new hardware-oriented approach for the extraction of disparity maps from stereo images. The proposed method is based on the herein named Adaptive Census Transform that exploits adaptive support weights during the image transformation; the adaptively weighted sum of SADs is then used as the dissimilarity metric. Quality tests show that the proposed method reaches significantly better accuracy than alternative hardware-oriented approaches. To demonstrate the practical hardware feasibility, a specific architecture has been designed and its implementation has been carried out using a single
FPGA chip. Such a VLSI implementation allows a frame rate up to 68 fps to be reached for 640 Â 480 stereo images, using just 80,000 slices and 32 RAM blocks of a Virtex6 chip.
Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction
Stereovision is a widespread sensing technique able to calculate depths of the objects in an observed scene by capturing and processing 2D images of the scene. As discussed by Nalpantidis and
Gasteratos [1,2], Munoz-Salinas et al. [3], Yu and Xu [4], the ability to reconstruct 3D information is crucial for a large variety of applications, such as robot navigation, surveillance, obstacle detection, autonomous vehicles, and many others.
A stereo vision system generally captures
Contents lists available at SciVerse ScienceDirect
Computer Vision and Image Understanding journal homepage: www.elsevier.com/locate/cviu
Adaptive Census Transform: A novel hardware-oriented stereovision algorithm q
Stefania Perri, Pasquale Corsonello ⇑, Giuseppe Cocorullo
Department of Electronics Computer Science and Systems, University of Calabria, Rende, Italy
a r t i c l e
i n f o
Article history:
Received 21 March 2012
Accepted 6 October 2012
Available online 17 October 2012
Keywords:
Stereovision algorithm
VLSI
Disparity Map Calculation
a b s t r a c t
This paper presents a new hardware-oriented approach for the extraction of disparity maps from stereo images. The proposed method is based on the herein named Adaptive Census Transform that exploits adaptive support weights during the image transformation; the adaptively weighted sum of SADs is then used as the dissimilarity metric. Quality tests show that the proposed method reaches significantly better accuracy than alternative hardware-oriented approaches. To demonstrate the practical hardware feasibility, a specific architecture has been designed and its implementation has been carried out using a single
FPGA chip. Such a VLSI implementation allows a frame rate up to 68 fps to be reached for 640 Â 480 stereo images, using just 80,000 slices and 32 RAM blocks of a Virtex6 chip.
Ó 2012 Elsevier Inc. All rights reserved.
1. Introduction
Stereovision is a widespread sensing technique able to calculate depths of the objects in an observed scene by capturing and processing 2D images of the scene. As discussed by Nalpantidis and
Gasteratos [1,2], Munoz-Salinas et al. [3], Yu and Xu [4], the ability to reconstruct 3D information is crucial for a large variety of applications, such as robot navigation, surveillance, obstacle detection, autonomous vehicles, and many others.
A stereo vision system generally captures
References: presence of non-ideal lighting, Image Vis. Comput. 28 (2010) 940–951. Syst. 58 (2010) 457–464. [5] O. Faugeras, Three-Dimensional Computer Vision: A Geometric Viewpoint, MIT Press, 1993. 20 (11) (2010) 1509–1519. Symposium on 3D Data Processing, Visualization and Transmission, Chapel Hill, North Carolina, USA, 2006, pp parallelism, J Real-Time Image Proc. 5 (2010) 291–304. FPGA-implementation, in: International Conference on Embedded Computer Systems, Samos, Greece, 2010, pp Intelligent Vehicles, San Diego, California, 2010, pp. 369–376. Cairo, Egypt, no. 3, 2003, pp. 1478–1481. Washington D.C., USA, no. 3, 2004, pp. 41. S. Perri et al. / Computer Vision and Image Understanding 117 (2013) 29–41 [18] M Image Underst. 114 (2010) 1180–1202. Vis. Image Underst. 114 (2010) 1303–1316. weight, IEEE Trans. Circuits Syst. Video Technol. 20 (6) (2010) 792–805. Technol. 20 (1) (2010) 15–26. (2010) 1643–1651. Artificial Intelligence, Sydney, Australia, 1991, pp. 1292–1298. Quebec, Canada, 2002, pp. 657–661. matching algorithm, Image Vis. Comput. 22 (12) (2004) 983–1005.