![](bg7.png)
7 DISCUSSION
Any image can contain both focused and unfocused
regions. There is no way to test the validity of a focus
detection algorithm without a reliable metric which
can quantify focus accuracy. By using the depth in-
formation from the light field we have shown that it is
possible to show that the focal response corresponds
to a specific image depth and that this depth changes
consistently with image focus.
Essentially, CDS highlights the parts of the image
that are in focus but also are more likely to contain
high-frequency information. As CDS creates new su-
perpixels on detecting image variation, there will be
some constant colour areas of the image that do not
change significantly with blurring. The result will be
that these regions are not marked as in focus. While
other methods can extract points of focus within the
image, these methods rely on tuning of the algorithm
parameters. As we have shown, the results on SML
can vary by as much as 50% depending on the se-
lection of adequate parameters. In addition, as these
are edge based techniques, they also rely on the ab-
sence of noisy edges in the image. CDS negates this
by considering regions within the image, as there is
an inherent averaging within ACWE.
This paper has described the first application of
superpixels in conjunction with scale-space. Apply-
ing CDS to the task of focus detection gives a result
which has been shown to correspond accurately to the
focal regions of the image. Crucially, it is unsuper-
vised and as such gives an unbiased representation of
the focus within an image, which can be demonstrated
by using the unique properties of Light Field Photog-
raphy.
REFERENCES
Adelson, E. H. and Wang, J. Y. A. (1992). Single lens
stereo with a plenoptic camera. IEEE Transactions on
Pattern Analysis and Machine Intelligence, 14(2):99–
106.
Bergholm, F. (1987). Edge Focusing. IEEE TPAMI,
9(6):726–741.
Chan, T., Sandberg, B., and Vese, L. (2000). Active
Contours without Edges for Vector-Valued Images.
Visual Communication and Image Representation,
11(2):130.
Chan, T. F. and Vese, L. A. (2001). Active Contours Without
Edges. IEEE Trans. Image Processing, 10(2).
Chen, M.-J. and Bovik, A. C. (2009). No-reference image
blur assessment using multiscale gradient. In Quality
of Multimedia Experience, 2009. QoMEx 2009. Inter-
national Workshop on, pages 70–74.
Kadir, T. and Brady, M. (2001). Saliency, Scale and Image
Description. IJCV, 45(2):83–105.
Kovacs, L. and Sziranyi, T. (2007). Focus area extraction
by blind deconvolution for defining regions of inter-
est. Pattern Analysis and Machine Intelligence, IEEE
Transactions on, 29(6):1080–1085.
Levin, A. (2007). Blind motion deblurring using image
statistics. Advances in Neural Information Processing
Systems, 19:841.
Levoy, M. and Hanrahan, P. (1996). Light field rendering.
In Proc. Conf. on Computer Graphics and Interactive
Techniques, pages 31–42.
Lindeberg, T. (1994). Scale-space theory: A basic tool for
analyzing structures at different scales. Journal of Ap-
plied Statistics, 21(1):225–270.
Liu, R., Li, Z., and Jia, J. (2008). Image partial blur de-
tection and classification. In Computer Vision and
Pattern Recognition, 2008. CVPR 2008. IEEE Con-
ference on, pages 1–8.
Lowe, R. and Nixon, M. (2011). Evolving Content-Driven
Superpixels for Accurate Image Representation. In
Proc. ISVC2011, volume 6938 of Lecture Notes in
Computer Science, pages 192–201. Springer-Verlag.
Nayar, S. K. and Nakagawa, Y. (1994). Shape from fo-
cus. Pattern Analysis and Machine Intelligence, IEEE
Transactions on, 16(8):824–831.
Ng, R., Levoy, M., Br
´
edif, M., Duval, G., Horowitz, M., and
Hanrahan, P. (2005). Light field photography with a
hand-held plenoptic camera. Computer Science Tech-
nical Report CSTR, 2.
Tai, Y.-W. and Brown, M. S. (2009). Single image defo-
cus map estimation using local contrast prior. In Im-
age Processing (ICIP), 2009 16th IEEE International
Conference on, pages 1797–1800.
Wilburn, B., Joshi, N., Vaish, V., Talvala, E. V., Antunez, E.,
Barth, A., Adams, A., Horowitz, M., and Levoy, M.
(2005). High performance imaging using large camera
arrays. ACM Transactions on Graphics, 24(3):765–
776.
Witkin, A. (1983). Scale-space filtering. Intl. Joint Conf.
Art. Intell., 2:1019–1022.
VISAPP2013-InternationalConferenceonComputerVisionTheoryandApplications
270