superresolution algorithms testalmalence.com/doc/superresolution-comparison/super... · 2016. 4....
TRANSCRIPT
-
Superresolution Algorithms Test
4 Algorithms compared: Almalence (PhotoAcute application) Robust super-resolution Normalized convolution Iterated back-projection
Test suite and ISO 12233 test pattern:http://lcavwww.epfl.ch/software/superresolution/
-
Algorithms characteristicsAlmalence Iterated back-
projectionRobust super-resolution
Normalized convolution
Non-Iterative + - - +
Imaging-device pre-tuned
+ - - -
Noise robustness + - - +
(with additional pass)
Speed Very fast( 100-400 Mul/Add per pixel)
Slow( >10000 Mul/Add per pixel)
Slow( >12000 Mul/Add per pixel)
Very slow(singular value decomposition at every pixel)
-
References Iterated back-projection. M. Irani and S. Peleg,
Improving resolution by image registration, GraphicalModels and Image Processing, 53:231-239, 1991.
Robust super-resolution. A. Zomet, A. Rav-Acha, and S. Peleg, Robust Super-Resolution, Proceedingsinternational conference on computer vision and patternrecognition (CVPR), 2001.
Normalized convolution. Tuan Q. Pham, Lucas J. vanVliet and Klamer Schutte, Robust Fusion of IrregularlySampled Data Using Adaptive Normalized Convolution, EURASIP Journal on Applied Signal Processing, Vol. 2006, Article ID 83268, 12 pages, 2006.
-
ISO 12233 Test Chart Sigma SD10 Foveon sensor 4 frames
-
Originalcrops
Almalence SR
-
RobustSuper-Resolution
NormalizedConvolution
IteratedBack-Projection
-
Street Scene Casio EX-F1 Moving clouds 6 Images
-
Originalcrops
Almalence SR
-
RobustSuper-Resolution
NormalizedConvolution
IteratedBack-Projection
-
Single frame crop
2x interpolated(bicubic)
-
Almalence SR
-
NormalizedConvolution
-
SummaryAmong tested algorithms, Almalence Super Resolution is the only one that:
Provides noticeable resolution increase with no annoying artifacts Can be tuned to specific imaging device for achieving the best results Is usable for real-life scenes (robust to noise) Can be used in mobile devices (fast, small memory footprint, 32-bit integer
implementation, optimized for ARM) Thoroughly tested in real life (used in photo applications for both desktop
computers and mobile phones, hundreds thousands users)
Implementations: High resolution zoom for mobile phones (BlessN900 application for Nokia
N900 device) Superresolution for DSLR cameras (PhotoAcute Studio application for
professional photographers)
-
Contactweb: www.almalence.come-mail: [email protected]
More information:Whitepapers, examples, tests:http://almalence.com/materials.htmlSuperresolution FAQ:http://photoacute.com/tech/superresolution_faq.htmlExample of Superresolution processing with PhotoAcute Studio:http://photoacute.com/studio/examples/mac_hddSuperresolution application high quality zoom on mobile phones:http://blessn900.comSuperresolution application higher resolution for DSLR cameras:http://photoacute.com/studio
© Almalence, Inc.
http://www.almalence.com/mailto:[email protected]://almalence.com/materials.htmlhttp://photoacute.com/tech/superresolution_faq.htmlhttp://photoacute.com/studio/examples/mac_hddhttp://blessn900.com/http://photoacute.com/studio
Superresolution Algorithms TestAlgorithms characteristicsReferencesSummary
/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 300 /GrayImageDepth -1 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 1200 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile () /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org) /PDFXTrapped /Unknown
/Description >>> setdistillerparams> setpagedevice