Intro

Hurray, you stumbled upon the website of Johan de Bock.
In the top menu you can navigate to my work related stuff.
Through the bottom menu you can check out the things I am passionate about and my hobby projects.

Connected Components

I have developed an extremely efficient algorithm to calculate the connected components in a binary image.
I have written about this technique in a paper titled "Fast and Memory Efficient 2-D Connected Components Using Linked Lists of Line Segments" which has been published in IEEE Transactions on Image Processing.

If you want to experiment with this algorithm, you can download 2dcc.zip which includes a binary for Windows and for Linux. It is a command line utility and can be used in the following way:
./2dcc image.png mode thresh_low thresh_high cc_neighborhood time_iterations auto_realloc
image.png : a greyscale png image
mode : 0 : creates an xml file that lists the regions
       1 : times connected components on binary image
       2 : times connected components on line segments
thresh_low thresh_high : pixels with thresh_low <= intensity <= thresh_high are selected
cc_neighborhood : neighborhood used for connected components, 4 or 8
time_iterations : number of iterations connected components is run (to get accurate timing results)
auto_realloc : array for regions is reallocated in blocks of auto_realloc*image_rows cells
Some timing results on the 8192x8192 sized image peppers-grey-8192.png for different processors:
thresh_high = 255, cc_neighborhood = 8, auto_realloc = 3
thresh_low #cc #lineseg mode c2d_t7400 ci7_720qm ci7_920
0          1   8192     1    71.5ms    36.2ms    34.0ms
64         457 60450    1    73.8ms    41.4ms    38.2ms
128        601 74697    1    72.5ms    44.6ms    41.0ms
192        833 51957    1    66.3ms    42.5ms    39.5ms
255        0   0        1    56.7ms    35.2ms    32.8ms
0          1   8192     2    0.3ms     0.3ms     0.2ms
64         457 60450    2    2.2ms     2.2ms     1.7ms
128        601 74697    2    2.7ms     2.7ms     2.1ms
192        833 51957    2    1.9ms     1.9ms     1.4ms
255        0   0        2    0.0ms     0.0ms     0.0ms
If you use the results of my algorithm and write about it, please refer to the paper.
If you have any questions about the algorithm, I would be happy to answer them.

Based on the same principles I have also developed a 3-d connected components algorithm. It can be used on large medical datasets or for material analysis. If you want to use this technique, you can download 3dcc.zip which also includes a binary for Windows and for Linux. It is again a command line utility and can be used in the following way:
./3dcc prefix slices mode thresh_low thresh_high cc_neighborhood time_iterations auto_realloc
prefix : set of greyscale png slices: prefix000000.png, prefix000001.png, ...
slices : number of slices
mode : 0 : creates an xml file that lists the regions
       1 : times connected components on binary volume
       2 : times connected components on line segments
thresh_low thresh_high : voxels with thresh_low <= intensity <= thresh_high are selected
cc_neighborhood : neighborhood used for connected components, 6 or 26
time_iterations : number of iterations connected components is run (to get accurate timing results)
auto_realloc : array for regions is reallocated in blocks of auto_realloc*rows*slices cells

Publications


  • Submitted: J. De Bock and W. Philips. Fast and memory efficient 2-d binary morphology using line segments. IEEE Transactions on Image Processing, pages xxxx–xxxx, 2011.
  • J. De Bock and W. Philips. Fast and memory efficient 2-d connected components using linked lists of line segments. IEEE Transactions on Image Processing, 19(12):3222–3231, 2010.
  • W. De Muynck, A.M. Ramirez, N. De Belie, J. De Bock, and W. Verstraete. Evaluation of anti-fouling strategies on aerated concrete by means of an accelerated algal growth test. In 2nd International Conference on Concrete Repair, Rehabilitation, and Retrofitting, pages 131–132, 2008.
  • D. Babin, J. De Bock, A. Pizurica, and W. Philips. The shortest path calculation between points of interest in 3-d mri images of blood vessels. In ProRISC, pages 295–298, 2008.
  • D. Hollevoet, P. De Smet, J. De Bock, and W. Philips. Towards automated forensic fracture matching of snap-off blade knives. In SPIE, Interferometry XIV: Applications, volume 7064, page 70640Q, 2008.
  • B. Goossens, J. De Bock, A. Pizurica, and W. Philips. Low-dose ct image denoising by locally adaptive wavelet domain estimation. In IEEE Benelux Chapter on Engineering in Medicine and Biology, EMBS, 2007.
  • N. Petrovic, A. Pizurica, J. De Bock, and W. Philips. Watershed data aggregation for mean-shift video segmentation. In SPIE, Applications of Digital Image Processing XXX, volume 6696, page 66962C, 2007.
  • J. De Bock and W. Philips. Line segment based watershed segmentation. In Advances in Computer Vision and Computer Graphics, Third International Conference on Computer Vision/Computer Graphics, MIRAGE, volume 4418 of Lecture Notes in Computer Science, pages 579–586. Springer, 2007
  • J. De Bock and W. Philips. Reduced memory watershed segmentation algorithm. In 7th FirW PhD Symposium, pages 021–021, 2006.
  • J. De Bock, R. Pires, P. De Smet, and W. Philips. A fast dynamic border linking algorithm for region merging. In Advanced Concepts for Intelligent Vision Systems, ACIVS, 8th international conference, volume 4179 of Lecture Notes in Computer Science, pages 232–241. Springer, 2006.
  • R. Morent, N. De Geyter, C. Leys, E. Vansteenkiste, J. De Bock, and W. Philips. Measuring the wicking behavior of textiles by the combination of a horizontal wicking experiment and image processing. Review of Scientific Instruments, 77(9):093502–093502–6, 2006.
  • J. De Bock, P. De Smet, and W. Philips. A fast sequential rainfalling watershed segmentation algorithm. In Advanced Concepts for Intelligent Vision Systems, ACIVS, 7th international conference, volume 3708 of Lecture Notes in Computer Science, pages 476–482. Springer, 2005.
  • P. De Smet, J. De Bock, and H. Quang Luong. On the reconstruction of strip-shredded documents. In 1st annual IEEE BENELUX/DSP Valley Signal Processing Symposium, SPS-DARTS, pages 185–188, 2005.
  • J. De Bock, P. De Smet, and W. Philips. Image segmentation using watersheds and normalized cuts. In SPIE, Vision Geometry XIII, volume 5675, pages 164–173, 2005.
  • R. Pires, P. De Smet, J. De Bock, and W. Philips. Image segmentation using thick-fluid watersheds. In SPIE, Image and Video Communications and Processing, volume 5685, pages 932–939, 2005.
  • P. De Smet, J. De Bock, and W. Philips. Semiautomatic reconstruction of strip-shredded documents. In SPIE, Image and Video Communications and Processing, volume 5685, pages 239–248, 2005.
  • J. De Bock, P. De Smet, and W. Philips. Watersheds and normalized cuts as basic tools for perceptual grouping. In ProRISC, pages 238–245, 2004.
  • J. De Bock, P. De Smet, W. Philips, and J. D’Haeyer. Constructing the topological solution of jigsaw puzzles. In IEEE International Conference on Image Processing, volume 3, pages 2127–2130, 2004.
  • J. De Bock and P. De Smet. Semi-automatic reconstruction of fragmented 2d objects. In 56th Annual Meeting of the American Academy of Forensic Sciences, pages 147–148, 2004.
  • P. De Smet, J. De Bock, and E. Corluy. Semi-automatic jigsaw puzzle reconstruction of fragmented documents. Forensic Science International, 136:29–30, 2003.
  • P. De Smet, J. De Bock, and E. Corluy. Computer vision techniques for semi-automatic reconstruction of ripped-up documents. In SPIE, Visual Information Processing XII, volume 5108, pages 189–197, 2003.
  • J. De Bock. Computer algorithms for solving jigsaw puzzles. Master’s thesis, Ghent University, 2003.

Techno

These are mixes made under my DJ pseudonym AlphaGamma.
Expect some deep dark minimalist techno.

































Foobar2000

Easily the best audio player available: foobar2000.
I have made a handy plugin for it that interfaces with Last.fm playlists and can handle XSPF shareable playlists.
You can find it here: Soft Playlists.

Fractals

These are pictures made with my Buddhabrot renderer.
More information can be found at BuddhaBrotMT.



Contact

Work email address: johan.debock@telin.ugent.be
Personal email address: johan.debock@gmail.com