CN103500442A - X-ray image multi-scale detail enhancement method in integrated circuit packaging - Google Patents

Abstract

The invention discloses an X-ray image multi-scale detail enhancement method in integrated circuit packaging, which comprises the steps of 1) acquiring X-ray images oriented to an integrated circuit packaging process; 2) performing Laplacian pyramid decomposition to the X-ray images to obtain X-ray pyramid detail images at all scales; 3) adjusting the brightness and the contrast of the images at the bottom layer of a pyramid by adopting a logarithm enhancement method; 4) adjusting the brightness and the contrast of the detail images at the top layer of the pyramid by adopting a histogram equalization enhancement method; 5) reconstructing the detail images at all scales to obtain detail enhanced images. The method disclosed by the invention has the advantages that the method is simple and fast, the dynamic range compression of images can be realized aiming at the X-ray images with low signal-to-noise ratio and poor defect contrast in the integrated circuit packaging, the useful information of the images is expanded, the image details are enabled to be more outstanding, the contrast of the images is improved and the accuracy is higher.

Description

The multiple dimensioned detail enhancing method of radioscopic image in integrated antenna package

Technical field

The present invention relates to a kind of detail enhancing method of radioscopic image, the multiple dimensioned detail enhancing method of radioscopic image in especially a kind of integrated antenna package.Belong to image processing field.

Background technology

Along with the development of IC (Integrated Circuit Design) encapsulation technology, components and parts inside is detected and proposed requirements at the higher level.In encapsulation process, the pad long term exposure is in air, easily oxidation, defect may appear: the crack of connection welding, do not have to connect, the cavity of solder joint too much, the problem of wire and wire pressure welding and nude film and linkage interface, the colloid that conducting resinl connects also can produce bubble in encapsulation process, and these all exert an influence to the package quality of great scale IC.And the surperficial invisible defects such as salient point connection, tie point rosin joint and silicon chip fine crack, colloid bubble all can't judge by the AOI technology.X ray, because of the transmission performance of itself, can detect surperficial invisible defect, so the X ray detection obtains applying more and more widely.

Yet, because the encapsulation components and parts that detect are generally the high density material, larger to the uptake of X ray, the radioscopic image gray scale is on the low side; And the encapsulation components and parts mostly are the thin slice chip, defect part and non-defect part are very little to the uptake difference of X ray, and radioscopic image also has the characteristics of low contrast.Add the stochastic distribution of defect arbitrary shape, make defect extract difficulty and increase.Therefore need first to radioscopic image, strengthen processing, make the image after processing be more conducive to follow-up defects detection.

Existing digital image enhancement technology is divided into two kinds of spatial domain and frequency fields: the Enhancement Method based on spatial domain is directly to be treated to basis to image pixel, and the treatment technology based on frequency domain be the Fourier transform of revising image is basis, the problem of these two kinds of method maximums has no idea to distinguish useful information and garbage exactly.In addition, the method for multi-scale enhancement image also has fast wavelet transform, but wavelet transformation can bring artificial artifact when strengthening complicated structure.

In sum, for the problem existed in conventional images enhancing technology, need to there is new method to strengthen image detail information, finally reach outstanding image detail, provide the purpose of effective information.

Summary of the invention

The objective of the invention is in order to solve the defect of above-mentioned prior art, provide a kind of image detail that makes more outstanding and improve the multiple dimensioned detail enhancing method of radioscopic image in the integrated antenna package of contrast of image.

Purpose of the present invention can reach by taking following technical scheme:

The multiple dimensioned detail enhancing method of radioscopic image in integrated antenna package is characterized in that said method comprising the steps of:

1) gather the radioscopic image towards the integrated antenna package process;

2) radioscopic image is carried out to Laplacian pyramid, obtain the X ray pyramid detail pictures of each yardstick;

3) to bottom in pyramid, adopt the logarithm Enhancement Method to adjust the brightness and contrast of image;

4) the detail pictures histogram equalization Enhancement Method of top layer in pyramid is adjusted to the brightness and contrast of image;

5) detail pictures under each yardstick is rebuild and obtained the image that details strengthens.

As a kind of embodiment, step 2) described radioscopic image is carried out to Laplacian pyramid, specific as follows:

2.1) to original image, adopt low-pass filtering and down-sampling to obtain the approximate image of a thick yardstick, decompose the low pass approximate image obtained;

2.2) the low pass approximate image obtained is carried out to interpolation and filtering, then calculate the difference of low pass approximate image and original image, obtain the detail section of image;

2.3) at the low pass approximate image obtained, adopt low-pass filtering and down-sampling to carry out the next stage decomposition, obtain new low pass approximate image;

2.4) repeated execution of steps 2.2)～2.3), complete multiple dimensioned decomposition.

As a kind of embodiment, step 2) in, the laplacian pyramid of image generates one of existence and dwindles contrary expansion process with image, the l tomographic image is expanded to the process identical with the l-1 tomographic image:

A) the expansion operator Expand of definition image is:

${g_l}^{*}=\mathrm{Expand}\left(g_l\right)$

G _l ^*and g _l-1measure-alike, specifically by being carried out to interpolation amplification, the l tomographic image obtains:

${g_l}^{*}(i,j)=4\underset{m=-2}{\overset{2}{\mathrm{\Σ}}}\underset{n=-2}{\overset{2}{\mathrm{\Σ}}}w(m,n)g_l(2i+m,2j+n)$

In this formula, and if only if coordinate during for integer, set up;

B) laplacian pyramid l tomographic image may be defined as:

$b_l(i,j)=g_l(i,j)-4\underset{m=-2}{\overset{2}{\mathrm{\Σ}}}\underset{n=-2}{\overset{2}{\mathrm{\Σ}}}w(m,n)g_{l+1}(2i+m,2j+n)$

b _l(i，j)＝g _l(i，j)-Expand(g _l+1)

Complete laplacian pyramid is defined as follows:

$\left\{\begin{array}{cc}{b}_l(i,j)=g_l(i,j)-\mathrm{Expand}\left(g_{l+1}\right),& 0\&le;l<L\\ b_L(i,j)=g_N(i,j),& i=L\end{array}\right.$

Wherein, L is pyramidal total number of plies.

As a kind of embodiment, step 3) the described brightness and contrast who bottom in pyramid is adopted to logarithm Enhancement Method adjustment image, adopt following computing formula:

B _l(i，j)＝255/log(255)*log(b _l(i，j)+1)

Wherein, b _l(i, j) is l layer detail pictures, B _l(i, j) is the image after processing;

As a kind of embodiment, step 4) the described histogram equalization of the detail pictures to top layer in pyramid Enhancement Method adjusts the brightness and contrast of image, specific as follows:

4.1) number of times that occurs of each gray level of statistic histogram

$P_r\left(r_k\right)=\frac{n_k}{n}0\&le;r_k\&le;1,k=\mathrm{0,1}...,L-1$

Wherein, P _r(r _k) be the probability of k level gray level; The number that L is gray level; n _kfor the number of times of this gray level occurs in image; The sum that n is pixel in image;

4.2) the normalized histogram of accumulative total

$S_k=\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}\frac{n_j}{n}=\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}{P}_r\left(r_j\right),0\&le;r_k\&le;1,k=\mathrm{0,1},...,L-1$

Wherein, S _kfor the histogram equalization transforming function transformation function; n _jthe number of times of this gray level appears in image; The sum that n is pixel in image;

4.3) calculate new pixel value

$S_k=\frac{255}{n}\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}{n}_j.$

As a kind of embodiment, step 5) the described image that obtains the details enhancing that detail pictures under each yardstick is rebuild, specific as follows:

To the top b of laplacian pyramid _lcarry out Expand, acquired results with strengthen through histogram equalization and logarithm each layer of detail pictures addition of laplacian pyramid that algorithm is adjusted.

The present invention has following beneficial effect with respect to prior art:

The present invention has advantages of simple and fast, it is low for signal to noise ratio (S/N ratio), radioscopic image in the poor integrated antenna package of defect contrast, considered the characteristic of image itself, make the radioscopic image in integrated antenna package effectively be strengthened, can realize the compression of dynamic range of images, make image carry out the expansion of useful information, thereby make image detail more outstanding and improve the contrast of image, higher accuracy is arranged.

The accompanying drawing explanation

The schematic flow sheet that Fig. 1 is the inventive method.

Fig. 2 is that the inventive method adopts the structural drawing of Laplacian pyramid to image.

Fig. 3 a is the X ray original image that the inventive method obtains; Fig. 3 b is the image after the inventive method is processed the X ray original image.

Embodiment

Embodiment 1:

As shown in Figure 1, the multiple dimensioned detail enhancing method of the radioscopic image in the integrated antenna package of the present embodiment comprises the following steps:

1) gather the radioscopic image towards the integrated antenna package process, be untreated original image, as shown in Figure 3 a, and with matrix representation;

2) radioscopic image is carried out to Laplacian pyramid, obtain the X ray pyramid detail pictures of each yardstick, specific as follows:

2.1) to original image, adopt low-pass filtering and down-sampling to obtain the approximate image of a thick yardstick, decompose the low pass approximate image obtained;

2.2) the low pass approximate image obtained is carried out to interpolation and filtering, then calculate the difference of low pass approximate image and original image, obtain the detail section of image;

2.3) at the low pass approximate image obtained, adopt low-pass filtering and down-sampling to carry out the next stage decomposition, obtain new low pass approximate image;

2.4) repeated execution of steps 2.2)～2.3), complete multiple dimensioned decomposition.

Wherein, the laplacian pyramid of image generates one of existence and dwindles contrary expansion process with image, the l tomographic image is expanded to the process identical with the l-1 tomographic image:

A) the expansion operator Expand of definition image is:

${g_l}^{*}=\mathrm{Expand}\left(g_l\right)---\left(1\right)$

G _l ^*and g _l-1measure-alike, specifically by being carried out to interpolation amplification, the l tomographic image obtains:

${g_l}^{*}(i,j)=4\underset{m=-2}{\overset{2}{\mathrm{\&Sigma;}}}\underset{n=-2}{\overset{2}{\mathrm{\&Sigma;}}}w(m,n)g_l(2i+m,2j+n)---\left(2\right)$

In formula (2), and if only if coordinate

during for integer, set up;

B) laplacian pyramid l tomographic image may be defined as:

$b_l(i,j)=g_l(i,j)-4\underset{m=-2}{\overset{2}{\mathrm{\&Sigma;}}}\underset{n=-2}{\overset{2}{\mathrm{\&Sigma;}}}w(m,n)g_{l+1}(2i+m,2j+n)---\left(3\right)$

b _l(i，j)＝g _l(i，j)-Expand(g _l+1) (4)

Complete laplacian pyramid is defined as follows:

$\left\{\begin{array}{cc}{b}_l(i,j)=g_l(i,j)-\mathrm{Expand}\left(g_{l+1}\right),& 0\&le;l<L\\ b_L(i,j)=g_N(i,j),& i=L\end{array}\right.----\left(5\right)$

Wherein, L is pyramidal total number of plies, and the structure of Laplacian pyramid as shown in Figure 2.

3) to the detail pictures of bottom in pyramid, adopt the logarithm Enhancement Method to adjust the brightness and contrast of image, be shown below:

B _l(i，j)＝255/log(255)*log(b _l(i，j)) (6)

Wherein, b _l(i, j) is l layer detail pictures, B _l(i, j) is the image after processing.Consider that log (0) is minus infinity, meaningless for image, therefore formula (6) is revised as:

B _l(i，j)＝255/log(255)*log(b _l(i，j)+1) (7)

4) to the detail pictures of top layer in pyramid, adopt the histogram equalization Enhancement Method to adjust the brightness and contrast of image, specific as follows:

4.1) number of times that occurs of each gray level of statistic histogram

$P_r\left(r_k\right)=\frac{n_k}{n}0\&le;r_k\&le;1,k=\mathrm{0,1}...,L-1---\left(8\right)$

Wherein, P _r(r _k) be the probability of k level gray level; The number that L is gray level; n _kfor the number of times of this gray level occurs in image; The sum that n is pixel in image;

4.2) the normalized histogram of accumulative total

$S_k=\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}\frac{n_j}{n}=\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}{P}_r\left(r_j\right),0\&le;r_k\&le;1,k=\mathrm{0,1},...,L-1---\left(9\right)$

Wherein, S _kfor the histogram equalization transforming function transformation function; n _jthe number of times of this gray level appears in image; The sum that n is pixel in image;

4.3) calculate new pixel value

$S_k=\frac{255}{n}\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}{n}_j.---\left(10\right)$

Due to image represented image information difference under each yardstick, more approach the subimage at pyramid top, its detail section is more obvious, but because its tonal range is narrower, so adopt algorithm of histogram equalization to strengthen processing to it, expanded view is as grey level distribution; More approach the subimage of pyramid base, its basic details information is abundanter, but integral image is partially dark, so adopt logarithm to strengthen algorithm, it is processed, but integral image is partially bright.So just obtain the new laplacian pyramid image after conversion.

5) detail pictures under each yardstick is rebuild and obtained the image that details strengthens, specific as follows:

To the top b of laplacian pyramid _lcarry out Expand, acquired results with strengthen through histogram equalization and logarithm each layer of detail pictures addition of laplacian pyramid that algorithm is adjusted, as shown in Fig. 3 b, with the X ray original image shown in Fig. 3 a, compare, can obviously find out that the detail section of image is effectively strengthened, the contrast of image is improved.

The above; it is only patent optional embodiment of the present invention; but the protection domain of patent of the present invention is not limited to this; anyly be familiar with those skilled in the art in the disclosed scope of patent of the present invention; according to the present invention, the technical scheme of patent and patent of invention design thereof are equal to replacement or are changed, and all belong to the protection domain of patent of the present invention.

Claims (6)

1. the multiple dimensioned detail enhancing method of the radioscopic image in integrated antenna package is characterized in that said method comprising the steps of:

1) gather the radioscopic image towards the integrated antenna package process;

2) radioscopic image is carried out to Laplacian pyramid, obtain the X ray pyramid detail pictures of each yardstick;

3) to bottom in pyramid, adopt the logarithm Enhancement Method to adjust the brightness and contrast of image;

4) the detail pictures histogram equalization Enhancement Method of top layer in pyramid is adjusted to the brightness and contrast of image;

5) detail pictures under each yardstick is rebuild and obtained the image that details strengthens.

2. the multiple dimensioned detail enhancing method of radioscopic image in integrated antenna package according to claim 1, is characterized in that: step 2) described radioscopic image is carried out to Laplacian pyramid, specific as follows:

2.1) to original image, adopt low-pass filtering and down-sampling to obtain the approximate image of a thick yardstick, decompose the low pass approximate image obtained;

2.2) the low pass approximate image obtained is carried out to interpolation and filtering, then calculate the difference of low pass approximate image and original image, obtain the detail section of image;

2.3) at the low pass approximate image obtained, adopt low-pass filtering and down-sampling to carry out the next stage decomposition, obtain new low pass approximate image;

2.4) repeated execution of steps 2.2)～2.3), complete multiple dimensioned decomposition.

3. the multiple dimensioned detail enhancing method of radioscopic image in integrated antenna package according to claim 2, it is characterized in that: step 2) in, the laplacian pyramid of image generates one of existence and dwindles contrary expansion process with image, the l tomographic image is expanded to the process identical with the l-1 tomographic image:

A) the expansion operator Expand of definition image is:

${g_l}^{*}=\mathrm{Expand}\left(g_l\right)$

G _l ^*and g _l-1measure-alike, specifically by being carried out to interpolation amplification, the l tomographic image obtains:

${g_l}^{*}(i,j)=4\underset{m=-2}{\overset{2}{\mathrm{\&Sigma;}}}\underset{n=-2}{\overset{2}{\mathrm{\&Sigma;}}}w(m,n)g_l(2i+m,2j+n)$

In this formula, and if only if coordinate

during for integer, set up;

B) laplacian pyramid l tomographic image may be defined as:

$b_l(i,j)=g_l(i,j)-4\underset{m=-2}{\overset{2}{\mathrm{\&Sigma;}}}\underset{n=-2}{\overset{2}{\mathrm{\&Sigma;}}}w(m,n)g_{l+1}(2i+m,2j+n)$

b _l(i，j)＝g _l(i，j)-Expand(g _l+1)

Complete laplacian pyramid is defined as follows:

$\left\{\begin{array}{cc}{b}_l(i,j)=g_l(i,j)-\mathrm{Expand}\left(g_{l+1}\right),& 0\&le;l<L\\ b_L(i,j)=g_N(i,j),& i=L\end{array}\right.$

Wherein, L is pyramidal total number of plies.

4. the multiple dimensioned detail enhancing method of radioscopic image in integrated antenna package according to claim 1, it is characterized in that: step 3) the described brightness and contrast who bottom in pyramid is adopted to logarithm Enhancement Method adjustment image, adopt following computing formula:

B _l(i，j)＝255/log(255)*log(b _l(i，j)+1)

Wherein, b _l(i, j) is l layer detail pictures, B _l(i, j) is the image after processing.

5. the multiple dimensioned detail enhancing method of radioscopic image in integrated antenna package according to claim 1, it is characterized in that: step 4) the described histogram equalization of the detail pictures to top layer in pyramid Enhancement Method adjusts the brightness and contrast of image, specific as follows:

4.1) number of times that occurs of each gray level of statistic histogram

$P_r\left(r_k\right)=\frac{n_k}{n}0\&le;r_k\&le;1,k=\mathrm{0,1}...,L-1$

Wherein, P _r(r _k) be the probability of k level gray level; The number that L is gray level; n _kfor the number of times of this gray level occurs in image; The sum that n is pixel in image;

4.2) the normalized histogram of accumulative total

$S_k=\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}\frac{n_j}{n}=\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}{P}_r\left(r_j\right),0\&le;r_k\&le;1,k=\mathrm{0,1},...,L-1$

Wherein, S _kfor the histogram equalization transforming function transformation function; n _jthe number of times of this gray level appears in image; The sum that n is pixel in image;

4.3) calculate new pixel value

$S_k=\frac{255}{n}\underset{j=0}{\overset{k}{\mathrm{\&Sigma;}}}{n}_j.$

6. the multiple dimensioned detail enhancing method of radioscopic image in integrated antenna package according to claim 3, is characterized in that: step 5) the described image that obtains the details enhancing that detail pictures under each yardstick is rebuild, specific as follows:

To the top b of laplacian pyramid _lcarry out Expand, acquired results with strengthen through histogram equalization and logarithm each layer of detail pictures addition of laplacian pyramid that algorithm is adjusted.

Images