JPH06331621A - Method for analysis of mineral by image processing - Google Patents

Abstract

PURPOSE:To provide a method for analysis of minerals that utilizes image processing to allow desired minerals to be taken out almost completely even with a decrease in the width of thresholds, enables simplification of operations for setting the thresholds, and can also make quick, accurate measurements. CONSTITUTION:A sample which contains minerals is consolidated, polished, and then photographed using a color TV camera connected to a reflecting microscope, and its images are subjected to contour retention smoothing processes for red, green and blue, and the images processed are synthesized to obtain a color image, and specific minerals are identified using data about hue, brightness and purity obtained from that image, and the data are binarized.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for analyzing minerals by image processing, and in particular, for consolidating and polishing a sample containing minerals, identifying the contained minerals by microscopic observation, and determining the ratio thereof. The present invention relates to a mineral analysis method by image processing.

[0002]

2. Description of the Related Art Knowing the types and proportions of minerals contained in natural ores and various industrial raw materials is very important when effectively utilizing them or treating them to increase their value. It helps. Conventionally, visual analysis by microscope observation has been performed as one of the methods for measuring the type and proportion of minerals. With this visual analysis, the sample is mixed with Bakelite, etc., compressed and solidified, then cut, and its cross-section is mirror-polished and observed under a microscope, and the size of the specific mineral is determined by the scale set in the eyepiece of the microscope. By reading the number, the proportion of the mineral in the sample is obtained. Therefore, in order to carry out such a measurement, a great deal of experience and highly skilled technology are required, and a shortage of human resources has been a problem. Further, in order to obtain an accurate measured value, it was necessary to perform a reading operation for a considerable number of fields of view, which took time and labor, and was inefficient.

In order to solve such a drawback, the present applicant has already proposed an automatic measuring method using an image analysis device (Japanese Patent Laid-Open No. 1-25943). The automatic measuring method of the proposed invention stores the color peculiar to minerals in the image analysis device,
The ratio of regions of the same color is automatically calculated to obtain the ratio of minerals corresponding to that color. More specifically, in the method of the proposed invention, the color of the mineral is divided into three elements of color of hue, brightness, and purity, each of which is converted into a numerical value, and an upper and lower threshold value is set for each numerical value. From the image of the surface of the sample (hereinafter referred to as the original image), only specific mineral particles are visualized to obtain the image.
The area of the graphic image can be easily measured by a computer, and the composition ratio of minerals in the sample can be calculated.

[0004]

However, even in the case of the same mineral, especially in the case of natural minerals, the color may change subtly due to a slight difference in composition or crystal structure.
And if there are other minerals of similar color,
It was very difficult to identify them. That is, for a mineral with a large change in color, the threshold widths of the above three elements must be limited to a wide range, and there are other minerals with similar colors that overlap the threshold width. If you do, it will be difficult to distinguish them. If you set a wide threshold value to completely extract a mineral with a large color change, another mineral with a similar color will also be extracted, and if you try to eliminate it and narrow the threshold width, The target mineral cannot be completely extracted. In the case of human visual measurement, in order to determine what kind of mineral the particle is, it is necessary to comprehensively judge not only the color but also the subtle texture characteristics of the mineral surface and the degree of color change. There is no problem. For the reasons described above, the automatic measurement by image processing has a drawback that its application range is narrow.

Furthermore, even if the same mineral is used,
The color changed subtly due to surface oxidation after polishing, and especially when a large number of minerals with similar colors were present at the same time, it was necessary to precisely reset the threshold value for each measurement. Setting the threshold was the most laborious task of the measurement, and also required complete knowledge of minerals, making complete automation impossible.
The necessity of repeating such a troublesome work for each measurement significantly impairs the convenience of automatic measurement.

Therefore, the object of the present invention is that the target mineral can be taken out almost completely even if the width of the threshold value is narrowed, the operation for setting the threshold value can be simplified, and quick and accurate measurement is possible. The object of the present invention is to provide a mineral analysis method by simple image processing.

[0007]

In order to achieve the above-mentioned object, the present invention consolidates and polishes a sample containing minerals,
Photographed with a color television camera connected to a reflection microscope, the image was analyzed by an image analysis device, and the contours were saved and smoothed for each color of red, green, and blue, and the processed images were combined to obtain a color image, A mineral analysis method by image processing characterized by identifying and binarizing a specific mineral from the data of hue, brightness, and purity from the image is adopted.

[0008]

The sample of the present invention may be in the form of powder, particles or lumps, but since it is photographed by a reflection microscope, it is necessary to mirror-polish a smooth cross section. In the case of powder or granules, a sample is prepared by consolidating with a resin such as Bakelite, cutting and polishing by a known technique.

The reflection microscope, television camera, and image analysis device used in the present invention may be those commercially available in general, but are limited to those capable of color processing due to the nature of the method. The image analysis device is limited to a device capable of contour-preserving smoothing processing, and for example, Luzex (trade name) manufactured by Nireco Corporation can be used.

In the present invention, the surface of the sample is photographed by a color television camera connected to a reflection microscope, and the image is processed by an image processing device. The magnification of the reflection microscope is arbitrarily adjusted according to the size of the mineral to be measured, but if the magnification is too high, the number of fields of view that must be measured to obtain the required accuracy results increases, and the magnification is too low. And the resolution of the processor is insufficient and the measurement becomes inaccurate. Based on the experience of the present inventors,
It is preferable to adjust the magnification so that the diameter of the measurement target mineral particles is within 1/100 to 1/5 of the visual field range.

Originally, the photographed original image is a composite of monochrome images of primary colors of red, green and blue. In the present invention, the contour-preserving smoothing process is performed on each of the three monochrome images, and then these are combined again to obtain a color image. The contour-preserving smoothing process is a known image processing method and is a combination of a contour extracting filter and a smoothing filter. By this contour preservation smoothing processing, the boundary line where the color of the grain boundary surface is rapidly changed is saved as the contour, and the continuous minute color changes in the area inside the contour are averaged (smoothed). To a single uniform color. The identification and binarization of minerals by image analysis are performed by digitizing the color with respect to the three elements of brightness, hue, and purity. This contour preservation smoothing process averages subtle changes in the color of minerals, The threshold width can be limited to a narrow range.

In the prior art, in order to binarize a mineral having a large color range, a wide threshold value must be set, and minerals having a similar color may be taken in at the same time. there were. In addition, the color of the same mineral changed depending on the sample, so the threshold had to be corrected each time the measurement was performed.

On the other hand, in the method of the present invention, even in a mineral having a wide color range, the color is averaged, so that the threshold value can be easily set, and another mineral having a similar color is included. However, it is possible to identify with a rough setting. Furthermore, the change in color due to the measurement conditions is relatively small, and it is not necessary to readjust each time measurement is performed.

[0014]

【Example】

(Example) Next, an example of the present invention will be described. Copper concentrate was crushed to a particle size of 5 μm, mixed with Bakelite, compressed and solidified, and a cross section cut by a diamond cutter was mirror-polished to apply the method of the present invention to binarize constituent minerals. It was

First, an appropriate white light was applied to the sample surface, and the sample surface was photographed with a television camera connected to a microscope. At this time, the magnification of the microscope was adjusted so that the photographing area of the television camera was 400 μm in the horizontal direction. The photographed image was captured by an image analyzer (Luzex 5000S manufactured by Nireco Corporation) connected to a television camera and stored as a direct image. This direct image is a composite of red, green, and blue monochrome images. Contour preservation smoothing processing was performed on each of the three monochrome images, and the processed images were stored in another memory. . The processed image was recombined to obtain a color image, and the image was subjected to IHP conversion to obtain luminance, hue, and purity data for each pixel. The IHP conversion and the subsequent mineral identification method were performed in the same manner as in the previously proposed invention. Upper and lower threshold values were set for the brightness, hue, and purity data for each pixel, and only specific color portions, that is, target mineral particles were taken out. The threshold value at this time is a numerical value that serves as a reference when identifying the target mineral.

Table 1 shows the results of setting thresholds of brightness, purity, and hue for pyrite, chalcocite, chalcopyrite, and chalcopyrite, which are the main constituent minerals of the sample. Numerical values in the table are brightness indexes for brightness, where 0 is black and 255 is white, and for hue is an index that indicates the range of colors, 0 and 255 are red, 100 is green, and 150 is blue. In the meantime, they are those intermediate colors, and in the case of purity, 0 is gray and 255 is a primary color. The comparison table on the right side of Table 1 is a combination in which threshold values do not overlap for each element, that is, a combination that allows complete identification is marked with O. One of the three elements
Since it is possible to discriminate even if there is no threshold overlap, it can be seen from this table that all minerals can be discriminated from other minerals. For example, although it is most difficult to distinguish from other minerals, it is possible to distinguish it from the content of chalcopyrite by its brightness, that of chalcopyrite by its hue, and that of chalcopyrite by its purity.
Also, if the range of threshold values required for binarizing minerals becomes narrow, by setting a threshold value with a certain margin added to this, the color will be subtly changed due to the polishing state and surface oxidation. Even in a changing sample, the same threshold value can always be used for identification and binarization, and the measurement operation can be greatly saved.

Comparative Example As a comparative example, Table 2 shows the measurement results by the method of the conventional example. It can be seen that the threshold range is wide for any element and any mineral as compared with the examples. For this reason, the threshold overlap becomes large and it is only pyrite and chalcopyrite in brightness that are clearly identifiable. When binarization is performed in such a state, the figures of the respective minerals overlap each other, making it impossible to accurately measure the ratio.

[0018]

As is apparent from the above, according to the present invention, the threshold value can be set narrowly, the operation of setting the threshold value can be simplified, and even if the color is partially changed, it can be quickly and accurately determined. It is possible to automatically measure minerals by simple image analysis.

[Table 1]

[Table 2]

Claims (1)

[Claims] 1. A sample containing minerals is consolidated and polished, and then photographed by a color television camera connected to a reflection microscope, and the image is smoothed and contour-preserved for each color of red, green and blue by an image analyzer. Mineral analysis by image processing, characterized in that the processed images are combined to obtain a color image, and specific minerals are identified and binarized based on the hue, brightness, and purity data from the images. Method.