Available online at www.sciencedirect.com

ProcediaEngineering

Procedia Engineering 00 (2010) 000–000

www.elsevier.com/locate/procedia

2010 Symposium on Security Detection and Information Processing

Image noise reduction based on superposition algorithm used in X-

ray imager

Liang Wang , Xiaofeng Jin, Yunjian Fei

Beijing Heweiyongtai Sci & Tech Co.Ltd, Shijingshan District,BeiJing,100043,China

Abstract

Now, the X-ray detection technique is used popularly in security inspection. using this technique, shadows of guns or knives are

showed clearly to the inspector. Besides, this technique can be controlled easily, and it is safe to people generally. Because of

these, the X-ray technique is more and more important in security inspection[1]. When the X-ray penetrates several objects with

different thickness and different density, we will get a gray image from imager. The nonuniformity and discontinuity of X-ray

will bring noise, this noise seems to be grain, it will damage image and make inspector uncomfortable. The paper mainly

introduces two superposition algorithms to process this noise but do not damage the original image, and the inclusions tell us the

process is very effective.

Keywords: X-ray, Superposition algorithm, Image process, Noise

1. The theory of X-ray image system

X-ray can be used in nondestructive inspection because of its penetrability[4], besides, it has any other

characteristics, such as fluorescence and photography[2]. The different object has different density and different

absorbing capacity of X-ray, and transmission energy is also different, so we can get images with different gray[5].

Fig 1. X-ray imager elements frame

Corresponding author. Tel.: 010-51712630; fax: 010-51712717.

E-mail address: king@heweigroup.com

c⃝ 2010 Published by Elsevier Ltd.

Procedia Engineering 7 (2010) 286–289

www.elsevier.com/locate/procedia

1877-7058 c⃝ 2010 Published by Elsevier Ltd.doi:10.1016/j.proeng.2010.11.046

Liang Wang, Xiaofeng Jin, Yunjian Fei / Procedia Engineering 00 (2010) 000–000

2. Introduce the noise and ways of noise reduction

There are several reasons for noise generation. The first one, X-ray is nonuniform; the second one, screen’s

luminous reflectance and its smoothness are nonuniform; the last one, data transmission may bring image noise. In

the paper, we mainly process the noise generated because light intensity is not enough. The noise is random, the

image will be fuzzy if we use median filter algorithm[3] to reduce the noise, so we choose superposition algorithm

to reduce the noise. The fig 2 shows the X-ray image with noise to you.

Fig 2. Take a picture by X-ray technique

3. Algorithm implementation inVC

3.1 Three points’ superposition

This algorithm superposes three points in corresponding three images, images’ names are image1.bmp,

image2.bmp and image3.bmp, the image processed is saved as image4.bmp. The variable image(data) points to the

image’s pixel, so the calculation is:

image4(data)=image1(data)/3+image2(data)/3+image3(data)/3.

The corresponding code in VC is:

First , find three images you want to process, and define the path of image4.bmp.

CString strSavePath=strPath+"temp1.bmp";

CString strSavePath1=strPath+"temp2.bmp";

CString strSavePath2=strPath+"temp3.bmp";

CString strSavePath3=strPath+"temp4.bmp";

Second, read these images, and save their pixels in variable imgdata1 imgdata2 and imgdata3.

Img1.Read(strSavePath1);

Imgdata1=Img1.m_pImgData;

Img2.Read(strSavePath2);

Imgdata2=Img2.m_pImgData;

Img3.Read(strSavePath3);

Imgdata3=Img3.m_pImgData;

In the end, begin to calculate, get the pixel of a image, and save it in array m[] generate image4.bmp file.

L. Wang et al. / Procedia Engineering 7 (2010) 286–289 287

Liang Wang, Xiaofeng Jin, Yunjian Fei / Procedia Engineering 00 (2010) 000–000

for(i=0;i<921600;i++)

{

m[i]=(imgdata1[i]/3)+(imgdata2[i]/3)+(imgdata3[i]/3);

}

Img4.Write(strSavePath4);

Fig 3. An original picture and the picture processed

3.2 twenty-seven points’ superposition

This algorithm superposes twenty-seven points in corresponding three images, these points include current point

and other 8 points around it in the same image and corresponding points in other two images. The fig 4 shows you

position of 27 points.

Fig 4. Position of the twenty-seven points

This algorithm uses more points to calculate, it is more effective than the first algorithm, and it can be used in

processing video, only video with moveless objects. The paper introduces a equation to enhance contrast:

)(xMky (1)

The variable x is the gray value of current point, is the average of a image’s gray, is the variance of current

point neighboring area, the neighborhood size is 33 . K is a coefficient. This equation is a common one, if the

contrast value is low( is low), the gray value after calculation will be improved, oppositely the gray value will be

reduced. This method would aggravate noise, so it is not suitable for us. We need to change the equation as:

Mkxky 21 )(

(2)

288 L. Wang et al. / Procedia Engineering 7 (2010) 286–289

Liang Wang, Xiaofeng Jin, Yunjian Fei / Procedia Engineering 00 (2010) 000–000

Fig 5. An original picture and the picture processed with 27 points’ superposition

4. Conclusions

The paper introduces two superposition algorithm used in image noise processing, the algorithm improves

images’ definition, and reduce noise which raises in dark section, then it supplies a technique for deeper research on

image noise. Now, this algorithm has been used in X-ray imager, in future, we need to do deeper research on video

noise processing.

References

[1] Wang T.W., Evans J.P.O. Sterescopic dual-energy x-ray imaging for target materials identification. IEEProc Vis Image Signal Process 2003;150(2):122-130.

[2] Shi XinHua. Improving Object Classification in X-ray Luggage Inspection[D], Doctor Dissertation, 2000:27-28.

[3] Kenneth R.Castleman. Digital Image Process. Electronic Industry publish; 2003.

[4] S.Cadeddu, D.Caredda, and M.Caria. Pharos. A Spectrometer -on-a-chip for digital radiology systems with

spectral detection. Nuclear Instruments and Methods in Physics Research A 2002; 478:367-371.

[5] Richard D.R. Macdonald. Design and Implementation of a Dual-Energy X-ray imaging system for Organic

Material Detection in an Airport Security Application. Machine Vision Application in Industrial Inspection 2001;

43(1):31-41.

L. Wang et al. / Procedia Engineering 7 (2010) 286–289 289