臺灣博碩士論文加值系統

For the past many years, imaging technology has been flourishing innovation
very much. There have developed a lot of multimedia imaging equipment, such
as, SLR camera, dynamic sport camera, smartphone photography, vehicle driving
record etc. Nowadays, some multimedia imaging equipment is also generally be-
gun to be widely used in public. Unfortunately, despite the advances in image
processing technology is improved, some of the issues have not been absolutely
solved. For example, attempting to take pictures in the high-speed moving
situation causes motion blur on the pictures. These potential challenges are worth
to be researched and improved.
In this thesis, there are two steps for deblurring the images quickly. The first
part is clustering the data, and the other part is deblurring the images in details.
At first, we would simulate the blurred situation with different acceleration and
display deblurring effects on the motion images. During the experiment, we will
set up the photographic equipment to record motion blurred images at the different
acceleration on sliding track. In order to facilitate analysis of the acceleration
value, we utilize the neural network technique to deal for noise reduction on the
acceleration data. Then, we also use K-means methodology to cluster the
acceleration data into some clusters. Furthermore, we would determine the
optimal point spread function in deconvolution operation.

Abstract ............................................................................................................. I
Ackonwledge .................................................................................................. II
List of Figures ................................................................................................ IV
List of Tables .................................................................................................. V
List of Abbreviations ...................................................................................... V
Chapter 1 Introduction ..................................................................................... 1
1.1 Research Background ................................................................................ 1
1.2 Research Motivation .................................................................................. 1
1.3 Research Goal ............................................................................................ 2
1.4 Thesis Structure ......................................................................................... 3
Chapter 2 Related Work .................................................................................. 4
2.1 Acceleration Sensor ................................................................................... 4
2.2 Artificial Neural Network .......................................................................... 8
2.3 Clustering: K-means .................................................................................. 9
2.4 Deblurring Function ................................................................................ 10
2.4.1 Wiener Filter (WNR) ............................................................................ 10
2.4.2 Regularized Filter (REG) ..................................................................... 11
2.4.3 Maximum Likelihood Estimation (MLE) ............................................ 12
Chapter 3 Proposed Method .......................................................................... 13
3.1 Collection of Acceleration Data ............................................................ 13
3.2 Clustering with K-means Methodology ................................................ 15
3.3 Deblurring .............................................................................................. 15
3.3.1 Wiener Filter ......................................................................................... 21
3.3.2 Regularized Filter ................................................................................. 21
3.3.3 Maximum Likelihood Estimation ........................................................ 22
Chapter 4 Experimental Results .................................................................... 18
Chapter 5 Conclusions and Future Work ...................................................... 27
Reference ....................................................................................................... 28

List of Figures
Fig. 1: Arduino 9 axis motion shield ............................................................... 4
Fig. 2: Arduino Uno Rev3 ............................................................................... 4
Fig. 3: The ongoing discrete Kalman filter cycle. ........................................... 7
Fig. 4: Experimental environment ................................................................. 16
Fig. 5: Diagram of motion blur degradation Model ...................................... 19
Fig. 6: Acceleration with 30,000 Hz ............................................................. 19
Fig. 7: Acceleration with 50,000 Hz ............................................................. 20
Fig. 8: Acceleration with 70,000 Hz. ............................................................ 26
Fig. 9: Blurred checkerboard image .............................................................. 23
Fig. 10: Blurred text image ............................................................................ 24
Fig. 11: Blurred simple text image ................................................................ 24

List of Tables
Table. 1: K-means clustering results of acceleration data. ............................ 21
Table. 2: 3 Deblurring Functions’ Time Cost. .............................................. 25
Table. 3: Comparison of Proposed Method and Blind Deconvolution. ....... 26

List of Abbreviations
ANNs Artificial Neural Networks
PSF Point Spread Function

Reference
[1] Y. Avrithis, Y. Kalantidis, E. Anagnostopoulos, and I.Z. Emiris, W0eb-
scale image clustering revisited, in Proceedings of the ICCV, 2015, pp. 1502-
1510.
[2] G. R. Ayers, and J. C. Dainty, Iterative blind deconvolution method and its
applications, Optics letters 13(7), 547-549, (1988).
[3] A. Babenko, and V. Lempitsky, Additive quantization for extreme vector
compression, in Proceedings of the CVPR, 2014, pp. 931-938.
[4] S. Bianco, L. Celona, and R. Schettini, “Robust smile detection using
convolutional neural networks,” Journal of Electronic Imaging, 2016, 25(6), pp.
063002-063007.
[5] T. F. Chan, and C. K. Wong. Total variation blind deconvolution, IEEE
Transactions on Image Processing 7(3), 370-375 (1998).
[6] Y. Chauvin, and D.E. Rumelhart, Backpropagation: Theory, Architectures,
and Applications, Psychology Press, 1995.
[7] S. Cho, and S. Lee. Fast motion deblurring, In TOG, volume 28, pp. 145,
ACM, 2009.
[8] A. Goldstein, and R. Fattal, Blur-kernel estimation from spectral
irregularities, in ECCV, pp. 622-635 (2012).
[9] H. Jégou, M. Douze, and C. Schmid, Product quantization for nearest
neighbor search, Trans. PAMI 33 (1) (2011) 117-128, doi:10. 1109/TPAMI. 2010,
57.
[10] R. E. Kalman, “A new approach to linear filteting and prediction
problems', ” Joural of Basic Engineering, Vol. 82, No. 1, pp. 35-45, (1960).
[11] D. Krishnan, T. Tay, and R. Fergus, “Blind deconvolution using a
normalized sparsity measure,” in CVPR, 2011 IEEE Conference on IEEE, 2011,
pp. 233-240.
[12] A. Levin, Y. Weiss, F. Durand, and W. T. Freeman, Efficient marginal
likelihood optimization in blind deconvolution. In IEEE Conference on CVPR,
pp. 2657-2664 (2011).
[13] A. Levin, Y. Weiss, F. Durand, and W. T. Freeman, Understanding blind
deconvolution algorithms, TPAMI, 33(12):2354-2367, 2011.
[14] S. Lloyd, Least squares quantization in PCM, IEEE Trans. Inf. Theory 28
(1982) 129-137, doi:10. 1109/TIT, 1982, 1056489.
[15] L. Mai, and F. Liu, Kernel fusion for better image deblurring, In: IEEE
Conference on CVPR, pp. 371-380 (2015).
[16] A. Mittal, R. Soundararajan, and A. C. Bovik, Making a completely blind
image quality analyzer, IEEE Signal Processing Letters 20(3), pp. 209-212
(2013).
[17] M. Muja, and D. G. Lowe, Scalable nearest neighbor algorithms for high
dimensional data, IEEE Trans. Pattern Anal. Mach. Intell, 36 (2014) 2227-2240,
doi:10. 1109/ TPAMI, 2014, 2321376.
[18] Q. Shan, J. Jia, and A. Agarwala, High-quality motion deblurring from a
single image, In TOG, volume 27, pp. 73. ACM, 2008.
[19] J. Shi, and J. Malik, Normalized cuts and image segmentation, Trans.
PAMI 22 (8) (2000) 888-905, doi:10. 1109/34. 868688.
[20] J. Sivic, and A. Zisserman, Video Google: a text retrieval approach to
object matching in videos, in Proceedings of the ICCV, 2003, pp. 1470-1477,
doi:10. 1109/ CCV. 2003, 1238663.
[21] J. Pan, Z. Hu, Z. Su, and M. H. Yang, “Deblurring text images via l0-
regularized intensity and gradient prior,” in Proceedings of the IEEE Conference
on CVPR, 2014, pp. 2901-2908.
[22] O. Whyte, J. Sivic, and A. Zisserman, “Deblurring shaken and partially
saturated images,” International journal of computer vision, vol. 110, no. 2, pp.
185-201, 2014.
[23] X. Wu, V. Kumar, J. R. Quinlan, J. Ghosh, Q. Yang, H. Motoda, G. J.
McLachlan, A. Ng, B. Liu, P. S. Yu, Z. H. Zhou, M. Steinbach, D. J. Hand, and
D. Steinberg, Top 10 algorithms in data mining, Knowl. Inf. Syst. 14(1) (2007) 1-
37, doi:10. 1007/s10115-007-0114-2.
[24] L. Xu, S. Zheng, and J. Jia, “Unnatural L0 sparse representation for natural
image deblurring,” in Proceedings of the IEEE Conference on CVPR, 2013, pp.
1107-1114.
[25] R. Yan, and L. Shao, ‘Blind image blur estimation via deep learning’, IEEE
Transactions on Image Processing, 2016, 25, (4), pp. 1910-1921.
[26] H. Zhang and L. Carin, Multi-shot imaging: joint alignment, deblurring and
resolution-enhancement, in CVPR, pp. 2925-2932, 2014.
[27] Y. Zhao, and G. Karypis, Empirical and theoretical comparisons of selected
criterion functions for document clustering, Mach. Learn. 55 (2004) 311-331,
doi:10. 1023/B:MACH, 0000027785, 44527, d6.
[28] L. Zhong, S. Cho, D. Metaxas, S. Paris, and J. Wang, Handling noise in
single image deblurring using directional filters, In IEEE Conference on CVPR,
pp. 612-619 (2013).