Image analysis – Image transformation or preprocessing – Changing the image coordinates
Reexamination Certificate
1999-10-07
2003-06-24
Do, Anh Hong (Department: 2624)
Image analysis
Image transformation or preprocessing
Changing the image coordinates
C382S289000, C382S290000, C250S559060
Reexamination Certificate
active
06584236
ABSTRACT:
This patent application claims priority based on a Japanese patent application, H10-321041 filed on Nov. 11, 1998, and H10-285783 filed on Oct. 7, 1998, the contents of which are incorporated herein by reference.
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to an image processing apparatus which recognizes the characteristics of a semiconductor wafer pattern image that is stored in a semiconductor wafer manufacturing process and the characteristics of the wire pattern image of a print substrate or the like.
2. Description of Related Art
In a manufacturing/inspection process of a print substrate or a semiconductor wafer, when a defect or a foreign object is generated on the surface of the substrate, the defect or foreign object is observed using a SEM (Scanning Electronic Microscope) or the like. The observation result is then stored as image data. Many of the stored images of defect have similar structures. When observing a wafer image, if one wishes to see similar images stored in the image file, the only available option is to search for similar images one after another through the image file based on one's memory. Moreover, the characteristics of the image data are difficult to represent using simple numerical values. Therefore, the characteristics of the image data are recognizable only to the person who has observed the image. Hence, it is not easy to share the image data with another person. There is an increasing demand for an image processing apparatus capable of searching for the image data of a semiconductor chip automatically and efficiently.
When a defective portion of a semiconductor chip is observed using a SEM or the like, it can be seen that a regular wire pattern is arranged in a prescribed direction, and an irregular pattern such as a foreign object and open/short patterns exists in the regular wire pattern. In order to analyze specific characteristics of the defect in detail, the observer rotates or magnifies or shrinks the image. If the background wire pattern containing the defect is rotated, the image processing apparatus cannot recognize that the rotated image is identical to the original image. Therefore, the image processing apparatus cannot search for the image data automatically. For this reason, the direction of the background wire pattern needs to be held constant, that is, the image needs to be standardized. The image can be standardized by specifying the direction of the given wire pattern, that is, the inclination of the wire pattern.
Conventionally, the inclination of the wire pattern of an image of a photograph of a print substrate or a semiconductor that has been photographed using a SEM or the like and input is specified primarily using a template matching technique.
FIG. 1
is a conceptual picture of a process for specifying the inclination of a wire pattern using a template matching technique. First, an edge detection process is performed for the original image as shown in FIG.
1
(
a
) to create a binary image as shown in FIG.
1
(
b
). An edge detection filter is an image processing device which detects positions at which the luminance of a gray scale image changes rapidly. For example, a Sobel operator or the like is used as an edge detection filter. The edge detection filter for gray scale images is explained in detail, for example, in Pattern Classification and Scene Analysis, by R. O. Duda and P. E. Hart, Wiley, 1971. The image shown in FIG.
1
(
b
) obtained using the edge detection filter is converted into a binary image as follows. The edge portions of defective spots and the wire pattern are represented by pixel value 1. All the other portions of the image shown in FIG.
1
(
b
) are represented by pixel value 0. Next, templates for comparing with the image in which edges have been detected are prepared.
FIGS.
1
(
c
) and (
d
) show examples of template images. The template images are also binary images. In these images also, the pixel values of the straight lines in FIGS.
1
(
c
) and (
d
) are 1 and the pixel values of all the other portions are 0. Straight line templates inclined in various directions are prepared to construct a correlation image of the edge detection image shown in FIG.
1
(
b
). That is, the correlation image of the edge detection image shown in FIG.
1
(
b
) is constructed by multiplying the pixel value of each pixel of the edge detection image with the pixel value of the corresponding pixel of the template image. As a result, only when both the pixel values of the pixels of the edge detection image and the pixel values of the pixels of the template image corresponding to the pixels of the edge detection image are non-zero, the pixel values of the corresponding pixels of the correlation image become non-zero. The pixel values of all the other pixels of the correlation image become zero.
As shown in FIG.
1
(
d
), when a figure exists at the same position and same direction on the template image as in the edge detection image, many pixels having non-zero pixel values appear on the correlation image as shown in FIG.
1
(
f
). Therefore, a threshold value is set to the number of pixels whose pixel values are non-zero on the correlation image. When the number of pixels whose pixel values are non-zero exceeds the threshold value, it is judged that the corresponding template figure exists. For example, eight straight lines having the same slope are detected from the image shown in FIG.
1
(
b
). In a semiconductor wafer image, multiple straight lines having the same slope form a wire pattern. Therefore, the slope of the wire pattern can be specified by the above-described process.
Since the conventional template matching method uses an edge detection filter for a pre-process, the image of the wafer pattern as the object of observation needs to be clear. For example, as in the case of FIG.
2
(
a
), if the edge portion of the wire pattern is unclear, the edge cannot be detected by the conventional edge detection filter as shown in FIG.
2
(
b
). Even if the template matching process is performed on the image shown in FIG.
2
(
b
), straight lines cannot be detected. Hence, the inclination of the wire pattern cannot be specified.
Moreover, the conventional template matching method is very time consuming. For example, the amount of operation required to detect an image having the image size of N pixels ×N pixels, the resolution for specifying the inclination of the wire pattern of 5° unit over the range between 0° and 180° will be shown in what follows. When a Sobel 3×3 operator and a threshold value process are used for the edge detection filter, both the multiplication and the addition require 3×3×N
2
steps, and the condition processing operation requires N
2
steps. Next, in the template matching process, (multiplication×N
2
+addition×N
2
+condition process)×(180÷5)×N
2
steps are required. In total, both the multiplication and the addition require 9N
2
(1+4N
2
) steps, and the condition processing operation requires 37N
2
steps. Since the number of steps the multiplication and the addition require are proportional to the 4-th power of the image size N, the processing time increases rapidly as the image size is increased.
SUMMARY OF THE INVENTION
Given these problems, it is an object of the present invention to provide an image processing apparatus capable of specifying the inclination of an object of detection even from an unclear image whose edge cannot be detected by the conventional template matching method. It is also an object of the present invention to provide an image processing apparatus capable of specifying at high speed the inclination of an object of detection even from a large image which takes a long length of time for the conventional template matching method to process. These objectives are achieved by a combination of characteristics described in the independent claims of the present invention. Moreover, the dependent claims of the present invention determine further advantageous
Ichikawa Masayoshi
Maruo Kazuyuki
Yamaguchi Takahiro
Advantest Corporation
Do Anh Hong
Rosenthal & Osha L.L.P.
LandOfFree
Image processing apparatus and method for image processing does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image processing apparatus and method for image processing, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing apparatus and method for image processing will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3149524