Research And Application Of 3d Model Segmentation Based On Region Growth Technique

The 3d model is a kind of multimedia data type after sound,image and video,which is widely used in network and industry.In recent years,the demand for creativity and entertainment has led to the development of digital media technology and the Internet,which has brought great changes to people's real life.The segmentation of 3d grid model is one of the hotspots in digital media technology.As early as the 1990 s,the regional growth technique was proposed for the segmentation of 3d models and proved to be an effective method.However,over-segmentation still exists in the process of segmentation.In this paper,the region growth technique is improved,which can effectively divide the triangle mesh model into some meaningful sub-modules.Firstly,this paper analyzes the geometric attribute normal vector of the triangle surface,studies the Dupin standard shape,the first basic form of the surface and the second basic form of the surface,and improves the traditional method of selecting the seed point of the region growth.In this paper,the triangle surface with higher flatness is selected as the seed point,and the variance of the normal vector of the triangle surface and its related grid surface is calculated.The numerical value of variance represents the flatness of the face slice.By selecting the smaller variance value,the problem of "sub-parts" after model segmentation is solved to make the model segmentation achieve the desired effect.Then,aiming at the problem of how to develop proper growth criterion in the area growth technique,this paper makes a detailed study on the area growth technique,and proposes the triangle model segmentation based on the combination of the cosine difference of the area weighted normal vector and the overall torsional degree of the subclass as the area growth criterion.Finally,compared with watershed algorithm and traditional region growing technology,the improved region growing technology is proved to be simple,efficient and easy to implement.Effectively solve the problem of over segmentation and time-consuming in the process of segmentation.