A tri-directional spotlight filter for object contour extraction

In this study, an object contour extraction using local structure information (LSI) captured by tri-directional spotlight (TDSL) filter is proposed. The proposed TDSL filter is designed by using three directional spotlight filters with overlapping areas in capturing local structure information around a discontinuity along object contour. The TDSL filter for object contour extraction is consisted of three steps. First, the TDSL filter is placed on an initial contour point to calculate the local structure information in determining and storing the edge candidates between any two pre-selected initial contour points. Next, edge points with weak evidence will be filtered out to obtain an object contour wideband. By performing a thinning approach on the object contour wideband image, a closed one-pixel-width object contour will be finally obtained. Experimental results of the proposed method for manmade and real medical images exhibit that the proposed TDSL filter can enhance the accuracy of contour detection and extraction much better than other existing methods for object contour extraction.

[1]  Demetri Terzopoulos,et al.  Snakes: Active contour models , 2004, International Journal of Computer Vision.

[2]  John F. Canny,et al.  A Computational Approach to Edge Detection , 1986, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  Doo-Sung Ahn,et al.  Object contour following task based on integrated information of vision and force sensor , 2007, 2007 International Conference on Control, Automation and Systems.

[4]  H ARITHA,et al.  A Boundary Detection in Medical Images using Edge Following Algorithm Based on Intensity Gradient and Texture Gradient Features , 2015 .

[5]  Wenxian Yang,et al.  User-Friendly Interactive Image Segmentation Through Unified Combinatorial User Inputs , 2010, IEEE Transactions on Image Processing.

[6]  Pong C. Yuen,et al.  A contour detection method: Initialization and contour model , 1999, Pattern Recognit. Lett..

[7]  Josef Kittler,et al.  Region growing: a new approach , 1998, IEEE Trans. Image Process..

[8]  Din-Yuen Chan,et al.  An improved automatic initial snaxel selection with corner tracing for object contour extraction in medical image , 2011, 2011 Visual Communications and Image Processing (VCIP).

[9]  Thomas Martin Deserno,et al.  Special issue: BVM 2009 Advances and recent developments in medical image computing , 2010, International Journal of Computer Assisted Radiology and Surgery.