[其它] An Interactive Augmented Reality Coloring Book

彬彬

797 主题	1 听众	1万积分

资深设计师

Rank: 7 Rank: 7 Rank: 7

纳金币: 5568
精华: 0

电梯直达

楼主

发表于 2011-12-21 09:30:04 |只看该作者 |倒序浏览

Adrian Clark1, Andreas Dünser1, Raphaël Grasset1,2

1The HIT Lab NZ, University of Canterbury, Christchurch, New Zealand

2Institute for Computer Graphics and Vision, Graz University of Technology, Austria

INTRODUCTION

Creating entertaining and educational books not only

requires providing visually stimulating content but also

means for students to interact, create, and express

themselves.

We present a new type of mixed-reality book experience,

which augments an educational coloring book with usergenerated

three dimensional content. In particular, we

introduce a new technique to engage users in actively

creating AR book content in real time using a coloring book

metaphor. In this first prototype, users can color the pages

of the book, and the system automatically recognizes the

pages and the various colors used by the users, and directly

maps the colored end results to virtual pop-up scenes and

3D models.

Our system is based on natural feature tracking and image

processing techniques that can be easily exploited for other

AR publishing applications.

SYSTEM DESIGN

A natural feature registration approach was chosen to

support the augmentation and modification of a physical

book. To allow modification of the content of the book, the

registration needed to be sufficiently robust to changing

targets. As even a small change in the original image can

cause registration failure, a series of filters were applied to

remove as much color as possible from each frame.

For each frame of video, the image contrast was

maximized while ensuring color constancy, and then white

balanced. The distance was calculated for each pixel from a

line between black and a predefined maximum gray value a

pixel can have which is still considered black. The final step

of color removal was to threshold all distances above a

given value to the color white. The final result is shown in

Figure 1.

Figure 1. Left: Original Image. Right: After Color Removal

Once the color has been removed from the frame, it can

be registered against known pages. To maximize robustness

and speed, the OPIRA registration library [1] was used in

conjunction with the SURF [2] feature detector.

When the black and white page is successfully registered,

the frame is rectified to the same orientation and dimensions

as the image of the page which was used for registration,

and the texture extracted. The page detection registration

process is stopped, and a new registration process for

tracking the rectified colored page is started.

EXAMPLE APPLICATION

The New Zealand Department of Conservation issues a

coloring book titled “Amazing Animals of New Zealand” to

primary school aged children to teach children about

animals native to New Zealand. Two pages of the book

were given to five children aged between 6 and 8 years old

to color in, and their colored pages were used as examples

for the application.

Color masks were created for each page, where every area

designed to pop up as a virtual image layer was colored in a

separate color. When the a frame is captured an alphablended

texture is created for each color. The background

texture is filled with the original image, such that when the

layers “pop-up” it appears the color has lifted off the page.

Every other texture is filled with the rectified image

captured at the end of the registration process.

Animated models were created for the most important

object in the scene, in the two examples, a Kiwi and a

Penguin. These models were texture mapped using the

original pages as the reference. When the page is detected,

the relevant area of the image is copied into the models

texture buffer, such that the model is now textured using the

image on the page. Example images are shown in Figure 2.

Figure 2. Screen shots from the Example Application

CONCLUSION

This work presents a new experience utilizing augmented

reality enhanced books where users are able to color in the

pages, and these pages are recognized by the system and

used to produce three dimensional scenes and textured

models reflecting the artwork created by the users.

The color removal system developed is able to

successfully allow automatic registration and texture

extraction from coloring book pages, even when the images

appear significantly different from the unmodified black and

white pages. The system was capable of running at real-time

frame rates, and with accuracy comparable to capturing the

images offline.

REFERENCES

[1] Clark, A., Green, R. and Grant, R.: 2008, “Perspective

correction for improved visual registration using natural

features”, Image and Vision Computing New Zealand,

2008. IVCNZ 2008. 23rd International Conference,

pages 1-6

[2] Bay, H.; Ess, A.; Tuytelaars, T.; Van Gool, L. "SURF:

Speeded Up Robust Features", Computer Vision and

Image Understanding (CVIU), Vol. 110, No. 3, pages

346-359, 2008