A Method of Spatial Segmentation of 3D Model Using Improved k-means

Abstract

The invention discloses a method of employing improved k-means to perform spatial domain segmentation on a three-dimensional model, comprising: first connecting the coordinates of vertexes of a curvature to be solved in all frames into a curve and employing a spectral analysis method to segment a target model; initially classifying the vertexes into blocks according to a vertex curvature expectation, i.e., classifying vertex with a same vertex curvature expectation into a block; respectively calculating the metric d between other vertexes and s vertexes; updating the center of each cluster, i.e., according to the expectation of all vertex metrics in each part, finding the vertex metric in the part which is proximate to the expectation to be used as a new clustering center of the part; and finally repeating until a threshold is smaller than a set value. The method employs a curvature to represent model motion intensity, and creatively defines the weighted mean between a Euclidean distance and a curvature expectation as a clustering metric; the clustering result guarantees space continuity as well as model motility.

Description

technical field [0001] The invention belongs to the multimedia technical field of three-dimensional animation model compression, and in particular relates to an improved k-means airspace segmentation method. Background technique [0002] With the continuous enrichment of 3D data acquisition methods, the maturity of computer graphics-related theories and technologies, and the rapid development of network technology, 3D models, as the fifth multimedia data type after text, audio, image (graphics) and video, are becoming more and more popular. Industrial manufacturing, product display, architectural design, robotics, medicine, e-commerce, education and training, military simulation, film and television entertainment and many other fields play an increasingly important role and give full play to its unique advantages. [0003] However, while increasingly sophisticated and perfect 3D models are widely used, the surge in data volume and complexity has brought great challenges to t...

AI Technical Summary

Problems solved by technology

But this algorithm has its inherent shortcomings. First, k in the algorithm is given in advance, and the selection of the value of k is difficult to estimate, because in most cases, it is not known how much a given data set should be divided into Secondly, the algorithm is very dependent on the selection of the initial value, and the algorithm often falls into a local minimum solution, and the results of different initial values ​​are often different; finally, the algorithm needs to continuously adjust the sample classification and continuously calculate and adjust After the new cluster center, so when the amount of data is very large, the time overhead of the algorithm is also very large

Images