دانلود رایگان مقاله ساماندهی مبتنی بر انتشار غیر یکنواخت برای تغییر شکل تغییرات

عنوان فارسی
ساماندهی مبتنی بر انتشار غیر یکنواخت برای تغییر شکل تغییرات
عنوان انگلیسی
Diffusion-based non-uniform regularization for variational shape deformation
صفحات مقاله فارسی
0
صفحات مقاله انگلیسی
9
سال انتشار
2016
نشریه
الزویر - Elsevier
فرمت مقاله انگلیسی
PDF
کد محصول
E513
رشته های مرتبط با این مقاله
مهندسی کامپیوتر
گرایش های مرتبط با این مقاله
طراحی کاربردی، الگوریتم ها و محاسبات
مجله
طراحی به کمک رایانه - Computer-Aided Design
دانشگاه
دانشکده علوم ساز و مهندسی، دانشگاه جنوب شرقی، چین
کلمات کلیدی
تنظیم، تغییر شکل تغییرات، انتشار
چکیده

Abstract


Regularization is a typical technique to correct the discontinuity artifacts at the control points in shape deformation. A regularizer with higher weights is required if the deformation is large, which will unfortunately distort the entire shape. In this work, we present a non-uniform regularization technique based on a shape-aware scalar field obtained from diffusion, which allows user to control the magnitude and range of the regularizer around specific control points. Experimental results show that shapes are deformed smoothly and no over-regularized artifact is observed with our non-uniform regularizer.

نتیجه گیری

5. Conclusion


In this work, we introduce a non-uniform regularization technique for variational shape deformation. We use diffusionbased techniques to generate a smooth scalar field and users are allowed to control the magnitude and range of the regularization. With our technique, it is possible to interactively control the deformation with no noticeable discontinuity artifacts, while the global shape distortion is kept lower. We extend the traditional regularization techniques to bring more flexible control of the regularizer. The flexibility is at the price that users are required to adjust more parameters. An interesting direction for future work is to automatically select parameters for any specified deformation. We approximate the diffusion equation with one linear equation in this work, which does not produce the exact solution on shapes. It is still worth studying how to efficiently solve the diffused weights with higher precision. Besides, our method works under the assumption that the smoothness of the distortion map around control points well follows the diffusion model. There is no theoretical guarantee about this assumption and it is essential to continue a theoretical study to model the discontinuity artifacts. In addition, since bijectivity is not explicitly modeled as a hard constraint in the deformation energy or the regularizer, the regularized deformation is not guaranteed to be bijective even if a strong quadratic regularizer is applied. We will keep exploring effective regularization techniques to suppress the inverted elements under extreme deformation.


بدون دیدگاه