3D modeling from sketches 

3D modeling from sketches introduction Sketching in 2D is much easier than in 3D because most designers can draw but are unfamiliar with 3D modeling tools. It is common to build a 3D model based on an existing object. But it takes professional skills and experience to develop an irregular shape. Even though we have these skills and experience, it still takes time and effort to draw and adjust the unique condition to meet our requirements. In particular, it will be much more practical to use a 2D shape that will come directly from an image, in the case where the object is of irregular shape. The object image can be a cartoon image or a sketched image.3d sketch

Extracting a 2D shape from an image remains a challenging task due to the endless possibilities of input sketch lines and different design techniques. Sometimes we need to segment the unique profiles into several smaller parts, making it easier to build models, but we need the shape as a whole for the same purpose. So it’s better to align the sketch lines to the image’s outline, rather than just showing the body extracted from the image based on the sketch lines. The exactness of the outcomes fibs in the correspondence between the hand-drawn sketch lines and the removed contour lines. Achieving the most accurate match requires not only interpreting the but of the sketch but also understanding the meaning of the contour lines extracted from the image. Both of these tasks are difficult for a machine to perform.

Related works

Our method is related to two types of modeling methods. One is based on pictures and the other on sketches. There are several image-based modeling methods and sketch-based modeling methods, each applied to different fins. We aim to realize a sketch-based modeling method with the information extracted from the image to exploit the intelligence of a back-end CAD application. Inspired by top-down alignment methods such as active contour models, we use alignment to replace artificial selection to get the most out of input images.

Image-based modeling

The system allows users to maintain a single view of the model.Chen et al.  Presented another image-based modeling technique called 3-sweep, which distorts input strokes based on the contour of the image.Presented a framework that allows users to participate in the entire design process of their things, from the initial design phase to the production of a new real-world object that fits well. . With the existing complementary objects.

Modeling from sketches

There are common problems in sketch-based modeling methods. In a 3D modeling system, view selection is critical to the modeling process. It is difficult and frustrating for users to define in multiple views. There is a solution in that allows the user to sketch in one view and deform the model in other pictures.  Generally, the sketch-based modeling method is based on sketch recognition techniques that convert a 2D data line into a 3D model like the systems.

 Active contour models

It uses an energy-minimizing spline guided by external stress forces and activated by image forces that fatigue it to features such as lines and edges. The initial contour of the traditional snake method must be close to the object’s boundary of interest in an image. Otherwise, it would converge to the wrong result. Additionally, active contours struggle to progress through boundary concavities. Also, the traditional snake method and its improved methods are likely to reach the local minimum because the snake patterns are not convex.

Overview

To build an image-based 3D model, our method needs to take an image with the target object as input. To design a 3D model, we first create each piece and then merge them into a single object. For the sketch-based modeling method, the sketch line allows the user to draw a part of the whole object. Figure 1(a) shows the user’s goal to decompose a raccoon’s head, while Figure 1(b) shows the intention to take the head as a whole object. In our method, the sketch line will line up with the contour lines of the background image to reveal the user’s original intent.