An MRF model-based approach to the detection of rectangular shape objects in color images

Rectangular shape object detection in color images is a critical step of many image recognition systems. However, there are few reports on this matter. In this paper, we proposed a hierarchical approach, which combines a global contour-based line segment detection algorithm and an Markov random field (MRF) model, to extract rectangular shape objects from real color images. Firstly, we use an elaborate edge detection algorithm to obtain image edge map and accurate edge pixel gradient information (magnitude and direction). Then line segments are extracted from the edge map and some neighboring parallel segments are merged into a single line segment. Finally all segments lying on the boundary of unknown rectangular shape objects are labeled via an MRF model built on line segments. Experimental results show that our method is robust in locating multiple rectangular shape objects simultaneously with respect to different size, orientation and color.

[1]  Martin D. Levine,et al.  Geometric Primitive Extraction Using a Genetic Algorithm , 1994, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  Josef Kittler,et al.  A survey of the hough transform , 1988, Comput. Vis. Graph. Image Process..

[3]  Ramakant Nevatia,et al.  Building Detection and Description from a Single Intensity Image , 1998, Comput. Vis. Image Underst..

[4]  Heikki Kälviäinen,et al.  An extension to the randomized hough transform exploiting connectivity , 1997, Pattern Recognit. Lett..

[5]  Ramakant Nevatia,et al.  Detection and Modeling of Buildings from Multiple Aerial Images , 2001, IEEE Trans. Pattern Anal. Mach. Intell..

[6]  Stanley M. Dunn,et al.  Approximating point-set images by line segments using a variation of the Hough transform , 1983, Comput. Vis. Graph. Image Process..

[7]  Hsien-Che Lee,et al.  Detecting boundaries in a vector field , 1991, IEEE Trans. Signal Process..

[8]  Evelyne Lutton,et al.  A genetic algorithm for the detection of 2D geometric primitives in images , 1994, Proceedings of 12th International Conference on Pattern Recognition.

[9]  Wen Gao,et al.  A robust method for unknown forms analysis , 1999, Proceedings of the Fifth International Conference on Document Analysis and Recognition. ICDAR '99 (Cat. No.PR00318).

[10]  Cláudio Rosito Jung,et al.  Rectangle detection based on a windowed Hough transform , 2004, Proceedings. 17th Brazilian Symposium on Computer Graphics and Image Processing.

[11]  W. Miller,et al.  Parallel implementation of the Hough transform for the extraction of rectangular objects , 1996, Proceedings of SOUTHEASTCON '96.

[12]  M. Levine,et al.  Extracting geometric primitives , 1993 .

[13]  Xiao Wang,et al.  Evolutionary optimization with Markov random field prior , 2004, IEEE Transactions on Evolutionary Computation.

[14]  Bridget Carragher,et al.  Automatic particle detection through efficient Hough transforms , 2003, IEEE Transactions on Medical Imaging.

[15]  Jianping Fan,et al.  Automatic image segmentation by integrating color-edge extraction and seeded region growing , 2001, IEEE Trans. Image Process..

[16]  Cormac Herley.  Recursive method to extract rectangular objects from scans , 2003, Proceedings 2003 International Conference on Image Processing (Cat. No.03CH37429).

[17]  P. Laguna,et al.  Signal Processing , 2002, Yearbook of Medical Informatics.