A Reseach To Feature-Enhanced Direct Volume Rendering

Direct volume rendering is one of the most effective methods for volumetric data visualization, while it usually needs a transfer function that assigns different optical properties such as color and opacity to the voxels. Consequently, it has been widely used in many fields of visualization technology, such as medical imaging and geological exploration. It is difficult to study how effectively visualize the feature information of the three-dimensional data field, while effective display features information of the volume data depends on the rendering and classification of volume data. Rendering is the process that maps three-dimensional spatial data to two-dimensional image, and the GPU ray-casting algorithm is the main method of currently direct volume rendering. Classification is mean to Construct the mapping relationship between the information of volumetric data and the optical properties, by analyzing and extracting features information in volume data, and thus demonstrate the interesting features. In the meantime, transfer function is common method for the classification of volume data. However, to find a good transfer function is time consuming, since it generally requires the considerable expertise of the user.In order to render plenty of features information in volume data, without needing complex transfer function, the paper did some exploration and research, and it proposes the Feature-based opacity modulation for direct volume rendering and feature enhanced volume rendering.(1) Feature-based opacity modulation for direct volume rendering. First, we use Moving least squares to reconstruct a continuous scalar curve of the ray profile. Then we obtain the feature of voxels from the ray profile of DVR, and these features are represented as different layers of sampling points. The opacity of these sampling points will be associated with the number of features of the ray. In order to achieve maximum visibility of the farthest sampling points, we modify the traditional ray integral for DVR. Local illumination is introduced to enhance shape perception of the features of the final rendering result.(2) Feature enhanced volume rendering. For enhancing important features of volume data, we proposed important weight function to weigh the importance of each feature in the volume data. While displaying all the features, enhancing rendering the high important weight features. In order to enhance visual and depth perception of rendered images, depth information of features and depth-based color cues are applied.The algorithm is implemented on the open source volume rendering framework-Voreen. Using QT, OpenGL, GLSL and other related technologies to integrate the algorithm into Voreen. Experiments with several volume data sets and our demonstration shows that our approach does not require the complicated transfer function and it can produce more features in the final rendered image than previous methods, such as DVR, MIDA, MIP and AVR.