Research On Transfer Function In Three-dimensional Cardiac Visualization

The virtual heart is a research topic, which represents heart tissues and structuresby mathematically modeling and computer simulation, which can be realized byThree Dimensional visualization(3D). The3D cardiac visualization is a process thatconstructs a3D heart model from a group of Two Dimensional (2D) cardiac imagesusing a series of operators such as volume rendering and presents it in the form of an2D image. The main role of transfer functions in the3D visualization is to maporiginal image information to optical properties. The performance of a3D cardiacvisualization is directly affected by transfer functions. Therefore, the study of transferfunctions plays an extremely important role in the research field of virtual heart.Firstly, this paper studies transfer functions for color images, and designstransfer functions that map multi-dimensional color information to opacity value.Most of the transfer functions map a scalar value like a gray value to opacity and fewof work defines opacity according to color information. By analyzing thecharacteristics of color space, this paper proposes a new algorithm that computesgradients using color distance. Then a Sobel operator is further adopted to improvethis algorithm. Finally transfer function is designed by using gradients to defineopacity to realize the function that extracts edge information of color images.The second aspect of study in this paper is about improving transfer functionsfor gray images. Calculating regional size value in the current transfer functionsrelies on the seed points selected by users, and regional size value is always affectedby the position of the seed point. To overcome these shortcomings, this paperconsiders calculating regional size using threshold. Histogram techniques are alsoadopted to convey more information to users. By improving histogram representation,each coordinate point in a histogram represents both the number of voxels and theirspatial location information. Finally the opacity is defined according to regional sizeand the parameters of the transfer functions are set interactively based on histogramrepresentation. Thus the function of visualizing similar regions in cardiac grayimages is realized.This paper designs and implements a virtual heart interactive visualizationsystem based on OpenGL and VTK. The experimental results are rather satisfactoryon testing the above algorithms on a variety of data sets. Experimental results showthat using color information to extract edge information reflects heart characteristicsbetter. The proposed transfer function based on regional size visualizes cardiacstructures better. Further, the adopted histograms ensure a realization of interactivetransfer function.