ISSN 2079-3537      

Scientific Visualization
Issue Year: 2017
Quarter: 4
Volume: 9
Number: 5
Pages: 105 - 116
Authors: N.I. Gavrilov (Russian Federation), E.P. Vasiliev (Russian Federation), I.V. Khramov (Russian Federation), A.A. Getmanskaya (Russian Federation), V.E. Turlapov (Russian Federation)
The paper is recommended by program committee of International Conference «Visual Analytics»
Address: N.I. Gavrilov
Lobachevsky State University, Russian Federation

E.P. Vasiliev
Lobachevsky State University, Russian Federation

I.V. Khramov
Lobachevsky State University, Russian Federation

A.A. Getmanskaya
Lobachevsky State University, Russian Federation

V.E. Turlapov
Lobachevsky State University, Russian Federation
Abstract: The goal of this work is the speedy incorporation of the cryo-imaging into visual analytics for practical medicine and biology, and creation a single solution for both managing visualization and segmenting the data of cryo-imaging in natural colors. In this paper we used the quantization and indexing of the natural colors for the cryo-imaging data as a basis for solving two formulated in the article problems: 1) three times more memory for cryo-imaging data, than for the same indexed data using palette; 2) the transparency control of real tissues which are heterogeneous in color.
We showed here that: 1) when using a palette of 256 colors, built using Linde-Buzo-Gray method, instead natural colors of femail dataset of Visible Human Project (VHP), the image quality losses are not significant (PSNR=44 - invisible for human eye); 2) 3D image quantization is not only the way to reduce dataset size – they showed about twice performance improvement (in frames per second, fps) compared with the visualization of three-channel raw data, and the best performance was achieved by loading the whole dataset as a single block (up to 108.9 fps for the dataset and GeForce GTX 680); 3) the high resolution of cryo-images makes rendering artifacts invisible to the human eye; after indexing the image quality is enough and trilinear interpolation becomes unnecessary, because it increase PSNR up to 45.5 db only, but programmable trilinear interpolation leads to twice downfalls the performance; the direct use of indices interpolation via hardware gives a the loss of the image quality which noticeable to eye; 4) quantization of natural colors give us good possibilities in the both opacity control and segmentation; a method for segmenting (classifying) a voxel tissue according to the normalized histogram of the indexed colors (NHIC-method) from its neighborhood is proposed and investigated (on the femail dataset from VHP); its possibilities were investigated in the range of both the neighborhood radius and the decision making threshold.
We plan to continue research in this direction and create algorithms and software to turn NHIC method into an open technology for cryo-imaging, biomedicine and education based on it.
Language: English

Open Article   Download ZIP archive