JPS60100032A - Automatic quantitative measurement for microscope image of sintered ore or the like - Google Patents

Abstract

PURPOSE:To enable an automatic quantitative measurement easily at a high accuracy with an image analysis high in the reproducibility by setting threshold for determining the concentration range of a composition to be measured based on a histogram depending on the concentration graduation of an image pickup signal of a microscope. CONSTITUTION:An image pickup signal of a sample through a microscope 2 and a TV camera 3 is processed with a binary-coding section 5 according to a number of thresholds from a threshold level generator 7 controlled with a synchronous separation part 6 via a amplifier 5. A binary-coded output according to the concentration graduation from the binary-coding section 5 is counted with a binary data counting section 9 via a gate responding to a mask area generating section 8 while the number of pixels constituting a screen area is counted with the generating section 8 and a mask area counting section 10. Then, the area ratio is calculated pertaining to graduations of concentration with a histogram data computing section 11 to automatically determine a reproducible histogram instantaneously with ease at a high accuracy. Referring he histogram thus obtained allows the setting of thresholds for classifying compositions different in the concentration thereby enabling an automatic quantitative measurement with an image analysis high in the reproducibility easily and at a high accuracy.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for automatically quantitatively measuring the composition of sintered ore in microscopic images.

Conventionally, as such an analysis method, after obtaining an ITV image of the tissue to be examined using a microscope, the examiner sets an appropriate threshold for compositional classification, and converts the target image into binary values using the threshold. A binary image of the desired tissue is extracted, and its area is calculated using an appropriate method. However, with this method of extracting and analyzing binary images related to each tissue, it is necessary to set and check the threshold each time in order to extract an appropriate binary image for each measurement field of view. ,
Moreover, there is a big problem in terms of reproducibility.

Therefore, there are many drawbacks such as poor measurement reproducibility, in other words poor accuracy, high skill required for the inspector, and inability to automate.

The purpose of the present invention is to eliminate such drawbacks, to enable highly accurate and highly reproducible image analysis without requiring any skill, and to provide automatic quantitative measurement of microscopic images of sintered ore, etc., which can be easily automated. The purpose is to provide a method.

To this end, in the present invention, a plurality of threshold levels are set in advance for classifying the degree of shading in an image into a large number of stages, and each of the unit area portions, that is, the pixels constituting the image is determined based on the image signal amplitude. The threshold level is sequentially converted into data indicating the density level, and based on these data, a histogram for each density level included in the image is instantaneously created by hardware, and the histogram thus obtained is The method is characterized in that by setting a threshold that determines the concentration range of each target composition to be measured based on , the area ratio and the like are calculated and measured using arbitrary software.

Embodiments of the present invention will be described below with reference to the accompanying drawings.

FIG. 1 is a block diagram schematically illustrating the method of the invention. For example, sintered ore is used as the sample 1, which is subjected to the appropriate surface inspection required for microstructural inspection using the microscope 2. The surface of the sample 1 to be inspected is imaged by the ITV 3 for each field of view enlarged by the microscope 2, and the image signal is given to the processing circuit. Here, the image signal is first transmitted to the amplifying section 4.
After being appropriately amplified, the signal is supplied to a binarization unit 5 provided for the processing featured in the method of the present invention. The image signal is also given by the synchronization separation section 6 as a synchronization signal to the threshold level generation section 7, and this threshold level generation section 7 sets the threshold level for binarization of the image signal in the binarization section 5 described above. It is designed to give A large number of threshold levels are preset in the threshold level generating section 7 to classify the light and shade in the image into as many stages as possible. Therefore, by using such a large number of threshold levels, the binarization section 5 synchronizes and sequentially converts the input image signal into binary data indicating one of the density levels based on its amplitude. For example, if the binarization unit 5 is configured to include an integrated circuit and performs conversion processing in 7 bits, 128 levels of binarization classification are possible, so the image signal can be thresholded based on its amplitude. Depending on the level, it can be obtained as binary data indicating any of 128 concentration levels. The binary data thus obtained is given to the binary data counting section 9 as binary data regarding a small area portion, that is, each pixel, forming the screen at a timing provided by the mask area generating section 8. This binary data counting section 9 counts the number of inputs for each density level, that is, the number of pixels of the same density. On the other hand, a mask area counting section 10 which receives the signal from the mask area generating section 8 counts the area of the screen, that is, the total number of pixels.

Concentration hist data calculation unit 11 into which these counted values are input.
For each density level, for example, 128 levels, the area ratio, that is, the ratio of the summed area of each density pixel to the total area of the visual field is calculated, and a gray histogram as shown in FIG. 2 is drawn. In particular, all these operations are performed by the circuit configuration, ie, hardware, as shown in the figure.
This instantly completes the required gray histogram. Since such arithmetic processing technology is well known, its explanation will be omitted. The calculation of the gray histogram as described above is similarly performed every time the field of view is moved by controlling the movement of the microscope 2. Further, such gray histogram data can be arbitrarily printed out by the printer 12, for example, as data for each visual field, or similarly printed out as average data for the entire visual field.

As described above, the gray histogram, which is a feature of the present invention, is used, and quantitative measurement of the composition based on the gray histogram will be further explained. First, by processing the initial visual field image using the hardware described above, a gray histogram as shown in FIG.
Suppose that it is printed out by . Although shown here as a curve graph, it is of course possible to use other graphs such as bar graphs or data displays. Although not mentioned above, it is of course possible for the examiner to view images of the visual field to be examined on a CRT screen in the conventional manner. The gray histogram obtained in this way has extremely high accuracy. Therefore, instead of simply setting thresholds corresponding to the composition based only on the CRT screen as in the past, inspectors can refer to the gray histogram, making it possible to set thresholds more accurately based on the transitions. . Thus, FIG. 3 shows, by way of example, four thresholds T+, TIT3. The case where Ta is set and compositions with five different concentrations are separated is shown. Of course, more classifications than this are also possible. In this way, the threshold 'r1. T,,T3. If T4 is set, since a gray histogram has already been obtained, the area ratios for each composition in this visual field image can be determined very easily.

These data are stored appropriately. Here, since each composition is scattered as minute parts, handling of these boundary parts is generally a problem, and conventionally known software for exclusion processing for the area near the boundary, that is, near the threshold, etc. Of course, it can be used as appropriate.

When the sample 1 or the microscope 2 is subsequently moved and the next field image is measured, a gray histogram for that image is obtained in the same manner as described above. At this time, even if the overall concentration level changes, it can be assumed that the relative concentration relationship of each component does not change. Therefore, for example, as shown in FIG. 3, each threshold T1 . 'rZ, T3. By storing the ratio of the Ta setting positions, it becomes possible to easily and automatically set appropriate thresholds for each composition in the image. Of course, in addition to this, it is also preferable to perform correction processing based on detection of peak positions, inflection positions, etc. in the gray histogram. In any case, due to the feature of the present invention of presetting the gray histogram, setting the required thresholds at the outset eliminates their resetting for subsequent field images, thereby speeding up and automating the measurements. becomes possible. In this way, measurements are taken for successive fields and a quantitative determination of each composition as a whole is achieved.

As described above, in the present invention, a large number of threshold levels are set in advance to determine density levels, and a gray histogram is created by determining the density level for each pixel that constitutes an image based on the amplitude of the ITV imaging signal. Then, tissue-specific thresholds are set based on the gray histogram to perform quantitative measurements of the desired composition, so the tissue-specific thresholds can be set at the beginning and can be omitted for successive fields of view, allowing for automation. becomes.

Furthermore, since the gray histogram can be quickly completed using hardware, the measurement time can be significantly reduced. Therefore, statistical measurement processing becomes possible. Furthermore, highly accurate measurements with reproducibility can be achieved without requiring the inspector to be highly skilled. In this way, extremely large effects can be obtained.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an outline of the present invention. FIG. 2 is a diagram showing an example of a gray histogram. FIG. 3 is a diagram showing threshold settings corresponding to measurement tissues regarding a gray histogram. 1. Sample 2.. Microscope 3.. ITV 4.. Amplification section 5.. Binarization section 6.. Synchronization separation section 7.. Threshold level generation section 8.. Mask area generation section 9.. Binary Data counting unit 10...Mask area counting unit 11...Historical data calculation unit 12...Printer patent applicant Japan Regulator Co., Ltd. Patent applicant
Nippon Steel Corporation Representative Patent Attorney Sakae Ono

Claims (1)

[Claims] This is a method of performing image analysis based on the density difference of each part of a microscope image, in which multiple threshold levels are set in advance to classify the degree of darkness in the image into a large number of stages, and the image is analyzed based on the image signal amplitude. Each unit area portion, that is, each pixel, constituting the image is sequentially converted into data indicating a density level using the threshold level, and based on these data, a histogram for each density level included in the image is instantaneously created by hardware. Therefore, by setting a threshold that defines the concentration range of each target composition to be measured based on the histogram obtained in this way, the area ratio etc. can be calculated and measured using arbitrary software. Automatic quantitative measurement method using microscopic images of sintered ore, etc.