Image Processing Reference
Comparison of DT-Binarize and max Limit
Max Limit DT-Binarize
This paper presents a new technique for image binarization problem using a group of different
thresholding methods. DT-Binarize is a supervised method with training and testing stages.
In the training stage, a decision tree is built using the standard deviation of the protein images.
Leaf nodes of the tree represent different thresholding techniques that provide the best binar-
ization method for a specific group of images. In the testing stage, using the decision tree, we
select the best thresholding technique for the test sample and then generate the binary image
using that technique.
We evaluate the performance of our approach with four different accuracy measures. For
all cases, DT-Binarize outperformed other single thresholding methods. Experimental results
show that our technique improves the binarization accuracy by 10% on the average and
provides high accuracy by reaching 95% of the expert choices.
 Zhu X, Sun S, Bern M. Classification of protein crystallization imagery. Engineering in
Medicine and Biology Society. 2004 IEMBS '04. 26th Annual International Conference of
the IEEE, vol. 1; 2004. p. 1628-31.
 Rupp B, Wang J. Predictive models for protein crystallization. Methods. 2004.
 Gonzalez R, Woods R. Digital image processing. Pearson/Prentice Hall; 2008. Available:
 Sezgin M, Sankur B. Survey over image thresholding techniques and quantitative
performance evaluation. J Electron Imaging. 2004. ;13(1):146-168. Available: ht-
 Ray N, Saha B. Edge sensitive variational image thresholding. IEEE International Con-
ference on Image Processing, ICIP 2007. 2007;vol. 6 p. VI37-VI40.
 Sigdel M, Pusey ML, Aygun RS. Real-time protein crystallization image acquisition
and classification system. Crystal Growth Design. 2013. ;13(7):2728-2736. Available: ht-
 Dinç I, Dinç S, Sigdel M, Sigdel MS, Pusey ML, Aygün RS. Dt-binarize: A hybrid bin-
arization method using decision tree for protein crystallization images. Proceedings of