Instead of a traditional method to observe these microstructures we use an additional video camera with a multistage digital image-processing system to analyze these microstructures quickly, powerfully and economically. This system consists of one set of CCD cameras and microscopes to scan the pattern, and is connected to the frame grabbers. The image also can be displayed simultaneously on the analog video monitor.

The frame grabber VIGAS is a plug-in card for the PC 586. It has a frame memory organized as 6404808 bits. The incoming video signal has a 8 bits resolution at a rate of 30 frames per second. An 8-bits A/D converter quantifies the signal to 256 gray levels. Each pixel in the frame memory may take a value between 0 and 255, with the value 0 corresponding to the black level and 255 to the white level according to a

modified look up table. We have notice the quantitative nonlinearity of the video signal and film thickness, so we modify look up table by deviding the image to ten circular regions and establishing ten transfer functions by experimental calibration. Fig.3(B) shows the histogram of a modified look up table will be more uniformly distribution than Fig.3(A), a unmodified one.

From the subtraction of a blank image and an image of the film we obtained the

image which concerns the thickness of a film. A higher gray level denotes a thicker emulsion or layer. We then Process the digital image and calculate the quality characteristics of the film.

In our optical system the value of K (a ratio of gray level to thickness) is not actually a constant for any value of thickness, But still be calculated as a constant without significant error. Although the function dependence of thickness and transmission is not linear but an exponential. But the non-linear part can be absorbed in the modified look up table. With the information of gray levels we define the new concepts of four film quality characteristics to evaluate the quality of the film of a holographic plate (for example, a light sensitive plate in dichromated gelatin (DCG) or photoresist emulsion). From the values of ATR; TPATR; MTR and RMSTR we can evaluate the

uniformity and quality of a holographic plate.

We can scan the full size of the plate to obtain the macro pattern. We also can select a magnification of 30, 200, 800, 1200 to obtain the local pattern. On the monitor screen we can see the microstructure of the surface of a film appear, and every part of a specimen must be put under consideration. But the higher resolution of the system has, the more measurement time to take. From the video system we can quickly analyze a 4"6" plate within 2 hours with our new software algorithms. Table 1 is one example of the measurement result of four film quality characteristics.

Fig.4 is an original image of a holographic plate with a particle on the micro-surface. After low pass filtering and thresholding the area and center of the shape of a particle is easy to calculate. When the area of the particle exceeds our tolerance value, a alarm signal will be yielding. By using Laplacian operation along the two axis direction we then obtain the length PL of the edge of a particle. From Fig.5 which zooms to a scale 1.5 : 1 we can obtain the distance of the short-axis and the maximum axis of the profile. The calculated values are listed as following:

thickness of film: 4.42 

area of the particle PA: 158 

length of the particle PL: 82 

The minimum axis : 2.5 

the maximum axis : 10.1 

Re: 4.12

Rc: 0.3

A comparison is made in Fig.6 by gauge measurement corresponding to the resist thickness of about 4.5 m. The gauge measurement of the thickness will be helpful for assisting judgment the thickness of a film in the film fabricating process. But it is a half-destructive testing and only used to obtain the relation of gray level and thickness. When the circularity Rc of a particle is larger than 0.7 or the ratio Re is smaller than 1.4 we will treat this particle as a dust, otherwise we think it's a chemical pollution spot on the film. Subsequently, Fig. 7 shows the classifier verification which selected the features, the ratio Re and circularity Rc as we define. We could find that each kinds of particles are well separated. These values will be helpful when we judge the impurities on the surface of a holographic plate.

nextindexback