2D/3D Image and Video processing

 

 

Members:

Prof. Dr. Emin Anarım

Asst. Prof. Burak Acar

Prof. Dr. Bülent Sankur

Assoc. Prof. Ayşın Ertüzün

Asst. Prof. Yücel Yemez (Koc University)

Prof. Lale Akarun (PRL)

Berk Gokberk (PRL)

Helin Dutağacı

Ceyhun Burak Akgül

Ender Konukoğlu

Erdem Yörük

Cem Demirkır

Hatice Çınar

Erhan Sütlü

 

 

Biometry and Man-Machine Interfaces

Biometrics addresses the problem of exploiting human body characteristics, such as face, voice, gait, fingerprint, hand geometry etc. to recognize people. We are also involved in the following projects:

Hand biometry

Researchers: Erdem Yörük, Bülent Sankur, Ender Konukoglu, Jerome Darbon

Description: To develop a hand-based biometric system using hand geometry and palm appearance data. A database of the hand images of about 500 people is made available on the web. 

 

Face detection and recognition

Researchers: Cem Demirkir, H. Kemal Ekenel, Bulent Sankur

Description: This work addresses the judicious combination of principal-component and independent-component features for person recognition. A wavelet-based scheme is investigated to combat the adverse effects of illumination and face expression. We also investigate algorithms for pose-independent face detection in cluttered scenes based on the idea of Viola-Jones and extend it to video tracking of faces.

 

Behavior understanding

Researchers: Hatice Cinar, Bulent Sankur

Description: We consider office environment where cameras track people and their  behavior is interpreted.

 

Publications:

E. Yorük, E. Konukoglu, J. Darbon, B. Sankur, Shape-Based Hand Recognition, IEEE Image Processing,  (under review),  2004.

H.K. Ekenel, B. Sankur, Feature selection in the independent component subspace for face recognition, Pattern Recognition Letters, 25, 1377–1388, 2004.

H.K. Ekenel, B. Sankur, Multiresolution Face Recognition, Image and Vision Computing, (under review), 2004.

V. Starovoitov, D. Samal, B. Sankur , Matching Of Faces in Camera Images and Document Photographs, ICASSP'2000, Int. Conference on Signal, Speech and Signal Processing, 5-9 June 2000, Istanbul.

M. Reijnders, P. van Beek, B. Sankur, J.C.A. van der Lubbe, Facial Feature Localization and Adaptation of a Generic Face Model for Model Based Coding, Signal Processing: Image Communication, vol. 7, 57-75, March 1995.

B. Esme, B. Sankur, E. Anarim, Facial Feature Extraction Using Genetic Algorithms, EUSIPCO-96, 8. European Signal Processing Conference, pp. 1511-1514, September 10-13, 1996, Trieste, Italy.

 

 

3D Image Processing

3D Archeological/Cultural Object Indexing

Researchers: Ceyhun Burak Akgul, Helin Dutagaci, Bulent Sankur

Description: 3D datasets are being prepared and novel modeling, registration, feature extraction, data fusion, and classification methods are considered.  We address specifically the determination of the class of museum art objects. This project is   cooperating with the Cost project entitled “Sculpteur”.

Info: http://www.sculpteurweb.org/html/

 

 

Video Coding

Our main interest has been in object-based video coding in the MPEG4 framework.  To this effect, we have developed video object tracking algorithms and metrics for the quality of tracked objects in the video.

 

Video object tracking

Researchers: Cigdem Erdem Eroglu, Bulent Sankur

Description: We have developed a scalable object-tracking framework, which is capable of tracking the contour of nonrigid objects in the presence of occlusion. The framework consists of open-loop boundary prediction and closed-loop boundary correction parts. The closed-loop boundary correction block employs a suitably weighted combination of low-level features such as color edge, color segmentation, motion models, and motion segmentation for each subcontour.

 

Image and Video Quality

Researchers: Cigdem Erden Eroglu, Ismail Avcibas, Bulent Sankur

Description: We determine measures to evaluate quantitatively the performance of video object segmentation and tracking methods without ground-truth (GT) segmentation maps, based on spatial differences of color and motion along the boundary and temporal differences between the color histograms. Canonical correlation analysis has indicated that ground-truthless measures perform as well as ground-truthed measures. We assess statistically several objective quality measures for still images with applications in steganalysis and coding.

 

Publications:

C. Erdem Eroglu, B. Sankur, M. Tekalp, Performance Measures for Video Object Segmentation and Tracking, IEEE Image Processing, 13(7), 937-950, July 2004.

C. Erdem Eroglu, B. Sankur, M. Tekalp, Video Object Tracking  with Feedbackof Performance Measures,  IEEE Circuit Systems and Video Processing, 13(4), 310-324, April 2003.

C.E. Eroglu, B. Sankur, Performance Measures for Video Object Segmentation and Tracking, SPIE Conf. Visual Communications and Image Processing 2003, Lugano, Switzerland, July 8-11 2003.

Avcibas, B. Sankur, K. Sayood, Statistical Evaluation of Image Quality Measures, Journal of Electronic Imaging, 11(2), 206-23, April 2002. (Availability software:  QUALITY for the calculation of several (26) image quality measures)

E. Anarım, Ç. Eroğlu Erdem, G. Karabulut, E. Yanmaz, , "Motion Estimation in the Frequency Domain Using Fuzzy C-Planes Clustering, IEEE  Trans. on Image Processing, October 2001.

 Y. Yemez, B. Sankur, E. Anarim, A Quadratic Motion-Based Object-Oriented Video Codec,  Signal Processing: Image  Communication, Vol. 15, pp. 729-766, 2000.

 

 

Image Compression and Segmentation

We have been involved in a comprehensive assessment effort for thresholding methods, especially in the nondestructive testing tasks. We have considered also color image segmentation schemes and the use of M-band wavelets.

 

Thresholding

Researchers: Mehmet Sezgin, Bulent Sankur

Description: We compare several (44) thresholding algorithms and develop an  evaluation platform. We also consider the use of thresholding in non-destructive testing environments. Multithresholding schemes are currently being assessed.

 

Color-image Segmentation

 Researchers: Fatih Kurugollu, Bulent Sankur:

 Description: We consider segmentation schemes for color images such as with multi-channel thresholding or constraint-satisfaction neural networks.

 

Compression

Researchers: Ismail Avcibas, Bulent Sankur

Description: We present a compression technique that provides progressive transmission as well as lossless and near-lossless compression in a single framework.

 

M-band wavelets

Researchers: Yucel Yemez, Emin Anarim, Bulent Sankur

Description:  We develop M-band wavelet schemes for the extraction of directional edges and for compression purposes.

 

Publications:

B. Sankur, M. Sezgin, A Survey Over Image Thresholding Techniques And Quantitative Performance Evaluation, Journal of Electronic Imaging, 13(1), 146-165, January, 2004. (Available software: OTIMEC with graphic user interface for the execution of several (42) image thresholding algorithms)

I. Avcibas, B. Sankur, N. Memon, K. Sayood, A Progressive Lossless/Near-Lossless Image Compression Algorithm, IEEE Trans. Communications, 2004.

M. Sezgin, B. Sankur, Image Multithresholding Based on Sample Moment Function, ICIP’2003: International Conference on Image Processing, Barcelona 2003.

F. Kurugollü, B. Sankur, E. Harmanci, Image Segmentation by Relaxation Using Constraint Satisfaction Neural Network, Journal of  Image and Vision Computing,  20, 483-497, 2002.

 I. Avcibas, N. Memon, B. Sankur, K. Sayood, Lossless and Near-Lossless Image Compression with Successive Refinement, IEEE Signal Processing Letters, 9(10), 312-314, October 2002.

V. Letournelle, B. Sankur, F. Pradeille, H. Maitre, Feature Extraction for Quality Asessment of Aerial Image Segmentation, ISPRS: International Soc. for Photogrammetry and Remote Sensing, PCV'02 Photogrammetric Computer Vision, 2002.

F. Kurugollü, B. Sankur, E. Harmanci Multiband Image Segmentation Using Histogram Multithresholding and Fusion,  Journal of  Image and Vision Computing,  9(13), 915-928, 2001.

H. Caglar, S. Güntürk, B. Sankur, E. Anarim, VQ-Adaptive Block Transform Coding of Images,  IEEE Trans. Image Processing, Vol. 7, pp. 110-116, January 1998.

Y. Yemez, T. Aydin, B. Sankur, E. Anarim, Multidirectional and Multiscale Edge Detection via M-Band Wavelet Transform,  IEEE Trans. Image Processing, Vol. 5, pp. 1370-1377, 1996.

 

 

Texture Analysis and Pattern Recognition

Researchers: R. Meylani, A. Latif-Amet, O.G. Sezer, A. Serdaroğlu A. Ercil,   A. Ertüzün

Description: To detect the defects encountered in textile images. Many different algorithms and methods are investigated for this purpose such as co-occurrence matrices, wavelets and subband domain analysis, 2-D lattice filters and subspace methods such as independent component and independent subspace analysis  methods.

 

Publications:

Serdaroğlu, A. Ertüzün and A. Ercil, “Defect Detection in Textile Fabric Images Using Wavelet Transforms And Independent Component Analysis” accepted, Pattern Recognition and Image Understanding : New Technologies, PRIA-7-2004,  Oct. 18-23, 2004, St. Petersburg, Russian Federation

O.G.  Sezer,  A. Ertüzün and A. Ercil, “Textile Fabric Defect Detection using Independent Component Analysis”, Submittd to Image and Vision Computing, 2003.

O. G. Sezer, A. Ertuzun and A. Ercil "Independent Component Analysis for Texture Defect Detection", Pattern Recognition and Image Analysis, vol. 14, no. 2, 2004,  pp.  303-307.

Latif Amet, A. Ertüzün and A. Erçil, ‘‘An Efficient Method for Texture Defect Detection: Subband Domain Co-Occurrence Matrices”, Image and Vision Computing, May 2000, pp.543-553.

S. Özdemir, A. Baykut, R. Meylani, A. Erçil and A. Ertüzün, "Comparative Evaluation of Texture Analysis Algorithms for Defect Inspection of Textile Products", in Proc. Int. Conf. on Computer Vision and Pattern Recognition (ICVP98), Aug. 14-17, 1998, Brisbane, Australia, pp.1738-1740.

Latif Amet, A. Ertüzün and A. Erçil, ‘‘Texture Defect Detection Using Subband Domain Co-occurrence Matrices’’, in Proc. IEEE Southeast Symp. for Image Analysis and Interpretation (SSIAI 98), April 6-7 1998, Tucson, Arizona, pp.205-210.

R. Meylani, A. Ertüzün and A. Erçil, "A Comparative Study on the Adaptive Lattice Filters for Defect Detection", in Proc. 3rd IEEE Int. Conf. on Electronics, Circuits and Systems (ICECS96), Oct. 13-16, 1996, Rhodes, Greece, pp. 976-979.

R. Meylani, A. Ertüzün and A. Erçil, "Texture Defect Detection Using the Adaptive Two-Dimensional Lattice Filter", in Proc. IEEE Int. Conf. on Image Processing, Sept. 16-19, 1996, Laussanne, Switzerland, pp. 165-168.

 

Theses