Graphics Reference
In-Depth Information
High-resolution makes the size of X-ray image is generally about 3000 * 1500. We
need a very large space to store images (the average size of each image is approx-
imately 2M bytes). And this can make it difficult to establish and maintain the large-
scale database. It would also require a larger computer memory and longer time. In
order to improve the effectiveness and efficiency of processing, we need to locate
tooth position from original X-ray images.
Tooth positioning refers to locate the area of tooth by a rectangular box from the
original X-ray image. There is not a separate subject for the research of the technology
of tooth positioning at now. Some proposed teeth recognition systems use the technol-
ogy complete artificial mark or artificial assisted semi-automatic mark to locate the
tooth areas. Nomir put forward an algorithm of artificial assisted semi-automatic for
tooth positioning. This method first mark the basic tooth areas manually, and then use
a probability model to calculate the overall position of the tooth. And it achieves the
position of each tooth precisely [3].
2
Algorithm of Tooth Positioning Based on Wavelet Transform
and Edge Detection
In this paper, we proposed the algorithm of tooth positioning based on wavelet trans-
form and edge detection. We get the feature patterns of both horizontal and vertical
direction by analyzing dental X-ray images. And then we test the feature patterns
respectively in both directions by using wavelet transform and edge detection so that
we get the line which can separate upper and lower jaws and the rectangle which lo-
cate the area of tooth.
2.1
Definition of Wavelet Transform
Continuous wavelet transform is proposed by the Morlet and Grossman [11], for the
function
, if its fourier transform
ˈ
(x
ˈ ˉ satisfies the equation (1), then
ˈ
(x
is
the wavelet function.
2
2
ˈˉ
()
ˈˉ
()
+
∞
0
d
d
C
ˈ
ˉ
=
ˉ
= < +∞
(1)
ˉ
ˉ
0
−∞
ˈ
()
t
In equation (1) ,
need to meet the requirements of the equation (2) in the domain
of time.
+∞
ˈ
t)dt
=
0
(2)
−∞
Take the continuous wavelet transform for the function
2
as shown in
ft
()
∈
LR
( )
equation (3)
Search WWH ::
Custom Search