NeRF

We present a method that achieves state-of-the-art results for synthesizing novel views of complex scenes by optimizing an underlying continuous volumetric scene function using a sparse set of input views. Our algorithm represents a scene using a fully connected (nonconvolutional) deep network, whose input is a single continuous 5D coordinate (spatial location ( x , y , z ) and viewing direction ( θ, ϕ )) and whose output is the volume density and view-dependent emitted radiance at that spatial location. We synthesize views by querying 5D coordinates along camera rays and use classic volume rendering techniques to project the output colors and densities into an image. Because volume rendering is naturally differentiable, the only input required to optimize our representation is a set of images with known camera poses. We describe how to effectively optimize neural radiance fields to render photorealistic novel views of scenes with complicated geometry and appearance, and demonstrate results that outperform prior work on neural rendering and view synthesis.

Transfer Functions in Volume Rendering of Lake Water Quality Considering Frequency Distribution

To explore the distribution and diffusion of pollutants in lakes, the volume rendering technique was used to express the lake water quality model in three dimensions. Due to the narrow distribution ranges and small spatial differences of the scalar field of the lake water quality mode, the perspective expression of subtle differences in the volume rendering process becomes important but difficult. In view of the foregoing case, this paper proposed transfer functions (TFs) in volume rendering of lake water quality considering the frequency distribution. The frequency distribution of the lake water quality scalar field was counted, the voxel ratios and the coloring probabilities of the frequency ranges were calculated, and then the voxel values were effectively mapped to colors and transparencies according to the coloring probabilities, to realize the refined expression of the differences in the spatial distribution of lake water quality. Experiments showed that the perspective expression of the subtle differences of lake (especially shallow lakes) water quality was improved using this method, which is conducive to analyzing the characteristics and changing laws of lake water quality model.

Effects of Different Parameter Settings for 3D Data Smoothing and Mesh Simplification on Near Real-Time 3D Reconstruction of High Resolution Bioceramic Bone Void Filling Medical Images

Three-dimensional reconstruction plays a vital role in assisting doctors and surgeons in diagnosing the healing progress of bone defects. Common three-dimensional reconstruction methods include surface and volume rendering. As the focus is on the shape of the bone, this study omits the volume rendering methods. Many improvements have been made to surface rendering methods like Marching Cubes and Marching Tetrahedra, but not many on working towards real-time or near real-time surface rendering for large medical images and studying the effects of different parameter settings for the improvements. Hence, this study attempts near real-time surface rendering for large medical images. Different parameter values are experimented on to study their effect on reconstruction accuracy, reconstruction and rendering time, and the number of vertices and faces. The proposed improvement involving three-dimensional data smoothing with convolution kernel Gaussian size 5 and mesh simplification reduction factor of 0.1 is the best parameter value combination for achieving a good balance between high reconstruction accuracy, low total execution time, and a low number of vertices and faces. It has successfully increased reconstruction accuracy by 0.0235%, decreased the total execution time by 69.81%, and decreased the number of vertices and faces by 86.57% and 86.61%, respectively.

Effects of Different Parameter Settings for 3D Data Smoothing and Mesh Simplification on Towards Near Real-Time 3D Reconstruction of High Resolution Bioceramic Bone Void Filling Medical Images

Three-dimensional reconstruction plays an important role in assisting doctors and surgeons in diagnosing bone defects’ healing progress. Common three-dimensional reconstruction methods include surface and volume rendering. As the focus is on the shape of the bone, volume rendering is omitted. Many improvements have been made on surface rendering methods like Marching Cubes and Marching Tetrahedra, but not many on working towards real-time or near real-time surface rendering for large medical images, and studying the effects of different parameter settings for the improvements. Hence, in this study, an attempt towards near real-time surface rendering for large medical images is made. Different parameter values are experimented on to study their effect on reconstruction accuracy, reconstruction and rendering time, and the number of vertices and faces. The proposed improvement involving three-dimensional data smoothing with convolution kernel Gaussian size 0.5 and mesh simplification reduction factor of 0.1, is the best parameter value combination for achieving a good balance between high reconstruction accuracy, low total execution time, and a low number of vertices and faces. It has successfully increased the reconstruction accuracy by 0.0235%, decreased the total execution time by 69.81%, and decreased the number of vertices and faces by 86.57% and 86.61% respectively.