A new localized phase congruency based object detector for X-ray images

In this paper, we present a novel method for automatically computing phase congruency at appropriate scales and orientations. Compared with Kovesi's phase congruency method, there are two distinct improvements in our method. First, our local phase information can be evaluated in a rotation invariant manner. Therefore, no orientation sampling is required. Second, the phase congruency is computed in a localized way, so there is no need to take all scales into consideration. Thus the computation load is reduced greatly and in the meanwhile this method can gain sub-pixel accuracy. We apply our algorithm to the endotracheal tube detection on X-ray image. Comparative experimental results show the efficiency of our method.

[1]  Tony Lindeberg,et al.  Edge Detection and Ridge Detection with Automatic Scale Selection , 1996, Proceedings CVPR IEEE Computer Society Conference on Computer Vision and Pattern Recognition.

[2]  Zheng Liu,et al.  Phase congruence measurement for image similarity assessment , 2007, Pattern Recognit. Lett..

[3]  Svetha Venkatesh,et al.  On the classification of image features , 1990, Pattern Recognit. Lett..

[4]  Ruud M. Snoeren,et al.  Image quality simulation and verification of x-ray volume imaging systems , 2006, SPIE Medical Imaging.

[5]  Tony Lindeberg,et al.  Feature Detection with Automatic Scale Selection , 1998, International Journal of Computer Vision.

[6]  David J. Fleet,et al.  Phase singularities in scale-space , 1991, Image Vis. Comput..

[7]  Peter Kovesi,et al.  Phase Congruency Detects Corners and Edges , 2003, DICTA.

[8]  Pengfei Shi,et al.  Ultrasonic Liver Discrimination Using 2-D Phase Congruency , 2006, IEEE Transactions on Biomedical Engineering.

[9]  Robyn A. Owens,et al.  Feature detection from local energy , 1987, Pattern Recognit. Lett..