Research On 3D-visualization Technologies Of Marine Environmental Data Field

Oceans play an important role in the political, economic and military fields. Mastery of the marine law will directly affect the economic ability and comprehensive national strength. As external manifestations of the marine law, marine environmental information has become a research hotspot at current domestic and foreign. Marine environmental information is often presented in the form of data field, and it is the key factor in using and understanding the oceans. By visualizing the marine environmental information, people are able to achieve a visual analysis of abstract data and stimulated in image thinking. Meanwhile, it highlights the meaning and essence of the research object. Thus, it is important to study and understand the oceans by visualizing the marine environmental information effectively.This paper conducts research around a number of key technics in marine environmental information visualization. Against the problems in data classification, data modeling and fast rendering, this paper proposes effective solutions and methods. The main works are as follows:(1) In the aspect of data clustering, this paper studies in the effective clustering methods and proposes a clustering algorithm based on the statistical characteristics that can realize the extraction of important areas. The algorithm realized the clustering of marine environmental data field by adopting fuzzy clustering method in a transform domain. First, the original data is mapped to the feature space by its statistical characteristics which lead to the data feature extraction and dimensionality reduction. Secondly, the exact surface of feature space is reconstructed by the Gaussian kernel function to eliminate the contaminated data points in the feature space. Finally, the FCM algorithm is introduced and modified for the clustering in marine environmental data field.(2) In the aspect of data modeling, this paper studies in the space segmentation modeling methods for volume rendering, and proposes the non-uniform spatial segmentation algorithm based on the voxel gradient by which the Non-uniform Spatial Segmentation Model(NSSM) is built for the realizing of ray-casting calculation. First, the voxel gradient is calculated in data field, the gradient histogram is constructed and the gradient threshold set is obtained. Secondly, the regions in data field are merged from bottom to top according to the gradient threshold set, and the non-uniform spatial segmentation model is built. Finally, the ray-casting calculation based on the model is advanced.(3) In the aspect of fast rendering, this paper studies in the parallel computing architecture and fast rendering methods, a fast ray-casting algorithm based on the CPU + GPU architecture is proposed to realize the fast volume rendering in massive data field. On the basis of parallel analysis to the NSSM based ray-casting algorithm, the intensive computationally part is transferred to GPU. Relying on the fast rendering algorithm and parallel computing hardware, the rendering efficiency is improved remarkably.