ICSM Group meetingHang XiaoOct 18, 2012
ICSM Organization
Regular Study Content: Books, tools … Each person is
responsible for one chapter
How many presenter each time ?
Paper Presentation Paper you are interested Implemented algorithm One main presenter At least one paper each
week
Hackathon Interesting algorithms Realize complex system
together C/C++, matlab, …
Questions Need one more meeting
each week? Extend the group?
Conflict with machine learning group.
Larger meeting room
Regular StudyBooks
《 Digital Image Processing 》
《 Machine Vision Theory, Algorithms, Practicalities 》
Tools ITK OpenCV
Paper Presentation Journals
Nature biotechnology, 23.3
Nature method, 19.276
PLOS Computational Biology, 5.515
Bioinformatics, 5.468
BMC Bioinformatics, 2.75
IEEE transactions on pattern analysis and machine intelligence (PAMI), 4.908
International Journal of Computer Vision , 3.741
IEEE transactions on image processing, 3.042
IEEE transactions on medical imaging, 3.643
IEEE transactions on signal processing, 2.628
Image and vision computing, 1.743
Signal processing, 1.567
Computerized medical imaging and graphics, 1.467
IEEE signal processing letters, 1.388
Journal of electronic imaging, 0.694
Image processing journals
Paper Presentation Journals
Pattern Recognition, 3.172
Journal of Machine Learning Research, 2.682
Data Mining and Knowledge Discovery, 1.545
…
Machine learning journals
Paper Presentation Conferences
CVPR: IEEE Computer Vision and Pattern Recognition
ICIP : IEEE Internal Conference on Image Processing
ICASSP: ICCV: International Conference on Computer
VisionACCV: Asian Conference on Computer Vision
Digital Image Processing
1236 - xiaoyong Frequency Domain Processing - xiaohang Wavelets - yangchen , wangcf Image Restoration - wangcf Image Compression - xiaoyong Morphological Image Processing --- Image Segmentation - xiaohang Object Recognition --- yangchen
Frequency Domain Processing
The 2-D Discrete Fourier TransformFiltering in the Frequency Domain
Wavelets
The Fast Wavelet Transform
Image Restoration Noise Model Restoration in the Presence of Noise Only – Spatial Filtering Periodic Noise Reduction by Frequency Domain Filtering Modeling the degradation Function Direct Inverse Filtering Wiener Filtering Constrained Least Squares (Reqularized) Filtering Iterative Nonlinear Restoration Using the Lucy-Richardson
Algorithm Blind Deconvolution Geometric Transformations and Image Registration
Image Compression
Coding Redundancy Interpixel RedundancyPsychovisual Redundancy JPEG Compression
Morphological Image Processing
Dilation and ErosionCombining Dilation and ErosionLabeling Connected ComponentsMorphological ReconstructionGray-Scale Morphology
Image Segmentation
Point, Line, and Edge DetectionLine Detection Using the Hough TransformThresholdingRegion-Based SegmentationSegmentation Using the Watershed Transform
Object Recognition
Computing Distance MeasuresRecognition Based on Decision-Theoretic
MethodsStructural Recognition
V3d-Convert
• Search• Abbreviation• History• 54291 line
• demo/demo.pro• demo/run_demo.h• demo/Run_demo.cpp
• Easy to use• Easy to write• Easy to manage• Easy to transfer
• 2d->3d convert• Algorithm collection /testing
Motivation Principle
AdvantageOrganization
Search
v3d_convert –list-help -list -search -history -component-tree-print -component-tree-create -component-tree-trim -component-tree-filter -component-tree-image -component-tree-dot -coseg-tree -coseg-image…~ 167 cmds
Search
v3d_convert –search swc –l 3 1 : -swc-adjust 2 : -swc-connect 3 : -swc-filter 4 : -swc-split 5 : -swc-stat 6 : -swc2marker 7 : -swc2svm 8 : -help bipartite-matching 9 : -help chanvese 10 : -help circle-forward… 29 : -help swc-filter 30 : -help swc-stat 31 : -help swc2marker 32 : -help swc2svm 33 : -help trace-between
Search
v3d_convert -fastmDo you mean -fastmarching
1: -fastmarching-linker 2: -fastmarching-mst 3: -fastmarching-phi 4: -fastmarching-tracing 5: -fastmarching-tree 6: -fastmarching-voronoi
choose : [1]
V3d-Convertsearch
Abbreviation
v3d_convert –topology-analysis rubin.tif –inswc ta.swc
v3d_convert –ta rubin.tif –is ta.swc
v3d_convert –tpa rubin.tif –is ta.swc
…
History
v3d_convert -history 1: v3d_convert -img 2: v3d_convert -img-erode 3: v3d_convert -img-erode 4: v3d_convert -img-erode 5: v3d_convert -search 6: v3d_convert -search good 7: v3d_convert -search filter 8: v3d_convert -search filter -l 3 9: v3d_convert -component-tree-filter 10: v3d_convert -mask
V3d-Convertabbreviation and history
V3d-Convert Organization Parameter Attribute {"+component-tree-create",0},{"-mb",1},{"-Mb",1},{"-ma",1},{"-outtree",1},{"-bin",0} Usage Func
int module_usage(BasicParser & parser){… if(cmd_name == "-component-tree-create") {
… }…}
Run Func int run_module(BasicParser & parser){… if(cmd_name == "-component-tree-create") {
… }…}
With Usagev3d_convert -component-tree-create
Usage :
v3d_convert <img_file> -component-tree-create [-mb <int>] [-Mb <int>] [-ma <int>] [-outtree <tree_file>]
create component tree and save to tree file
-mb minimum beta size-Mb maximum beta size-ma minimum alpha size
Without Usagev3d_convert -component-tree-create
Usage :
v3d_convert [<img_file>] -component-tree-create -mb <para> -Mb <para> -ma <para> -outtree <para> -bin
V3d-Convert and Vaa3D
Parameter Attribute {"+component-tree-create",0},{"-mb",1},{"-Mb",1},{"-ma",1},{"-outtree",1},{"-bin",0}
V3d-Convert and ITKv3d_convert –itk 1: -itk-AntiAliasBinaryImageFilter 2: -itk-BSplineWarping1 3: -itk-BSplineWarping2 4: -itk-BilateralImageFilter 5: -itk-BinaryMedianImageFilter 6: -itk-BinaryMinMaxCurvatureFlowImageFilter 7: -itk-BinaryThinningImageFilter (~149 itk cmds)
v3d_convert -itk-CannyEdgeDetectionImageFilterUsage: v3d_convert -itk-CannyEdgeDetectionImageFilter inputImage outputImage [variance upperThreshold lowerThreshold]
Thanks!