Research On The Method Of Muscle Fiber Texture Synthesis In The Virtual Surgery

The virtual surgery system is widely used in the medical field,it is the combination of computer graphics and virtual reality technology,it can provide doctors with operation plan and operation training,etc.There are more than 600 muscles in human body,muscle is an important organ of human body,it is a key point to improve the reality of muscles in the virtual surgery system.The traditional 2D texture synthesis is based on the similarity of images,which can not reflect the geometric features of the muscle surface,there are many problems for the traditional method,such as too much manual intervention and too long synthesis time.To solve above problems,in this paper,human skeletal muscles are taken as the research objects,a method of muscle fiber texture synthesis based on field-aligned parameterization is proposed,with the smoothest vector field or the alignment of the principal direction field as the guidance field of parameterization,it implements the control of the direction and proportion of muscle fiber texture,the geometric features are presented by automatic configuration of singularities,which improve the realism and real-time of texture synthesis.The main work and research results of this paper are as follows:(1)Construction of the smoothest vector field on the surface of muscle model.This paper presents a unified representation of the n-vector field by the n power of a complex number,a modified Dirichlet energy is proposed to obtain the smooth energy,the global smoothest vector field is obtained by minimizing the smooth energy,which can automatic configuration of singularities.When the energy equation is discretized,the problem is transformed into solving a generalized eigenvalue and its eigenvector,which solves by the Cholesky decomposition and the inverse power iteration method,so the construction of the smoothest vector field is completed.(2)Construction of the alignment of the principal direction field on the surface of muscle model.The product of the complex number and reference vector is used to represent a given direction,such as principal curvature directions.On the basis of the smooth energy,we add a linear term that describes the alignment of the direction field to construct the linear energy equation,which solves by the Cholesky decomposition,so the construction of the alignment of the principal direction field is completed.(3)The field-aligned parameterization of muscle model and muscle fiber texture synthesis.An improved method of field-aligned parameterization is proposed,with the smoothest vector field or the alignment of the principal direction field as the guidance field,which allows the additional singularities and omits unit-norm constraint and curl correction.This paper presents a method to solve singularities of the vector field by applying the branched coverings,the problem of minimizing Dirichlet energy is transformed into solving a generalized eigenvalue and its eigenvector,we use the inverse power iteration method to obtain the texture coordinates,the texture coordinates are adjusted to avoid the discontinuity of interpolation,then,we use OpenGL to render the muscle fiber texture.Finally,this paper design and implement a prototype system of the muscle fiber texture synthesis,the algorithm and visualization of each module are integrated,the experimental results and analyses are given,and the future works are presented.The experimental results show that we propose improved methods for the smoothest vector field,the alignment of the principal direction field,and the field-aligned parameterization,which are giving consideration to the reality of muscle fiber texture and the real-time performance of the virtual surgery system,the system also increases the interaction of muscle fiber texture synthesis.The methods not only have the significance for the muscle fiber texture in the virtual surgery,but also can be used as a reference for any directional texture synthesis,which have extensive application prospects.