GPU Based Computation of the Structural Tensor for Real-Time Figure Detection

In this paper we present a real-time realization of the method of detection of local structures in images of predefined orientation. The method is based on an analysis of the structural tensor computed in monochrome and color images. Thanks to the GPU implementation of the low-level feature detection an order-of-magnitude speed-up was achieved compared to the software implementation. The method can be used for real-time detection of solid objects in HDTV streams as shown by many examples.

[1]  Jack J. Purdum,et al.  C programming guide , 1983 .

[2]  B. Cyganek.  An Introduction to 3D Computer Vision Techniques and Algorithms , 2009 .

[3]  B. Jähne,et al.  A tensor approach for local structure analysis in multi-dimensional images , 1996 .

[4]  Johan Wiklund,et al.  Multidimensional Orientation Estimation with Applications to Texture Analysis and Optical Flow , 1991, IEEE Trans. Pattern Anal. Mach. Intell..

[5]  Silvano Di Zenzo,et al.  A note on the gradient of a multi-image , 1986, Comput. Vis. Graph. Image Process..

[6]  G. Clark,et al.  Reference , 2008 .

[7]  Boguslaw Cyganek,et al.  Adding parallelism to the hybrid image processing library in multi-threading and multi-core systems , 2011, 2011 IEEE 2nd International Conference on Networked Embedded Systems for Enterprise Applications.

[8]  Jitendra Malik,et al.  Blobworld: Image Segmentation Using Expectation-Maximization and Its Application to Image Querying , 2002, IEEE Trans. Pattern Anal. Mach. Intell..

[9]  Eero P. Simoncelli,et al.  Differentiation of discrete multidimensional signals , 2004, IEEE Transactions on Image Processing.

[10]  Jie Cheng,et al.  CUDA by Example: An Introduction to General-Purpose GPU Programming , 2010, Scalable Comput. Pract. Exp..