JPH0145663B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention is capable of analyzing the diffusion situation from photographic film taken of the spread of oil and pollutants spilled into the ocean, rivers, lakes, etc., and analyzing the combustion situation in furnaces, combustion chambers, etc. The present invention relates to a simple and rapid image processing method and apparatus for use in laboratories, production sites, etc.

Images, as a means for humans to obtain information, are superior to sounds and senses in terms of amount of information and processing speed. Recent developments in remote sensing technology, medical technology, and industrial technology have further increased the need for image processing, the functions and performance of image processing devices have become more diverse and sophisticated, and large-scale image processing systems have been developed. There is.

Image processing methods can be roughly divided into two types: analog methods such as photographs and videos, and digital methods mainly using computers. The analog method is
Although it has the characteristics of high speed, high density recording, and large screen,
There are problems with versatility, accuracy, adjustment, etc. on the other hand,
Digital methods can be combined with electronic computers to perform complex processing and can be repeated with high precision.
It has come to play a major role in image processing.

Conventional methods and systems for digital image processing are as follows. The system is mainly composed of a general-purpose computer, to which image input equipment and image output equipment specific to images are connected. Image input devices digitize images recorded on film or VTR for processing by electronic computers. Typical image input devices include flying spot scanners, drum scanners, laser beam scanners, and business computers. etc. are used.

As electronic computers, large electronic computers and minicomputers are used in terms of high-level language processing and high-speed calculations. Magnetic disks,
Must be equipped with magnetic tape. In addition, the main storage device is over 1MB, and the disk capacity is 200~
Requires 300MB. Software in an image processing system is comprised of routines for operating image input/output devices, arithmetic processing routines for processing image data given as a two-dimensional array, and image file manipulation and management routines. As an image output device, a television type display is used as a monitor. There are methods for recording image output, such as printing on a line printer, drawing on an automatic drafting machine, and drawing on an XY plotter.

With the above methods and processing devices, it is difficult to process the image data all at once in the main memory of a computer, and the image data must be stored in a file once and then sequentially processed as partial images. It takes more time than that. Since general-purpose computers perform sequential calculations, processing two-dimensional array data such as images requires an extremely long computer program and takes time. Also,
Software is created in conjunction with hardware, and its configuration varies and is not compatible.
It lacks practicality. The format of magnetic tape, which is the input medium for image data, is also inconsistent, and data processing is complicated.

As mentioned above, conventional methods and processing equipment are large-scale and expensive, and require control of temperature and humidity at the installation location, so they are not economical or simple to use in laboratories or production sites. It cannot be said that it is an advantageous method or device in terms of sex.

In order to eliminate these drawbacks, the present invention provides a method for analyzing oil and contaminants in laboratories and production sites.
This allows for simple and quick analysis of combustion conditions in furnaces and combustion chambers.

For this reason, in the present invention, the diffusion and combustion conditions of oil and pollutants are constantly photographed using a remote camera, and the continuous measurement of physical quantities such as thickness, concentration, and length at specific points in the gray scale image is carried out using a direct measurement method. After photoelectrically converting the photographed grayscale image, it is quantized as density level data using an analog-to-digital converter, the density level of each pixel is stored in a storage medium, and this data is read out and the density By drawing a distribution map using an X-Y block and correlating the density level near a specific point in the image with the physical quantity measured by the direct measurement method, the physical quantity for the image density can be determined, and the physical quantity distribution in the image can be determined. It is characterized by simple and quick measurement.

As an example of the present invention, a case will be described in detail in which the spread of oil spilled into a water tank is analyzed.

FIG. 1 shows an oil experiment tank. The size of the tank
The size of the tank is 15m x 15m x 0.4m, and the bottom of the tank is white. The oil used in the experiment was heavy oil A. A bay with a width of 30 cm and a depth of 50 cm was created in the center part 2 of one side of the tank, and 136 c.c. of heavy oil was continuously released from the tank every second. Photos were taken using color reversal film with a motor drive camera 3 installed at a height of approximately 14 m above the aquarium. In order to match the density of the oil on the photographic film with the thickness of the oil, oil thickness measuring instruments 4 were installed at the entrance of the bay and at two points 2 m from the entrance, which can measure the thickness of the oil with an accuracy of ±0.1 mm. Measure the thickness of the oil.

FIG. 2 is a block diagram of an image processing apparatus for analyzing the oil diffusion situation from the shading of a photographic film image. The photographic film 6 irradiated by the light source 5 is photographed by a solid-state imaging type television camera 7 serving as an image input device. As a TV camera, geometric distortion,
A solid-state imaging camera with little shading is used.

FIG. 3 shows a video signal output from the solid-state imaging television camera 7, which is a 1V _PP signal consisting of a video portion of 0.7V _PP and a synchronization signal of 0.3V _PP . This video signal is sent to an app-log digital conversion section 8 and converted into a digital signal. During digital conversion, the lower limit and upper limit of the video signal are arbitrarily set using a variable dial, and the value between these values is converted into a 4-bit digital signal at high speed. Since one screen is captured at 1/60 seconds, it can process not only still images but also videos.

The image memory 9 stores image data in which one screen is divided into 256 x 256 picture elements, and each picture element is digitally converted into 4 bits, and has a memory capacity of 32 Kbytes.
It is connected to the image processing device 11 via. The image arithmetic processing device 11 is composed of a microprocessor, but since the memory capacity of 32 Kbytes is insufficient for processing, a RAM memory 13 with a capacity of 256 Kbytes is connected via the input/output interface 12. During high-speed capture, up to 8 images can be stored. When processing films, etc., a large amount of data is stored on a 280 Kbyte floppy disk 14.

The image data stored in the image memory 9 is read out in synchronization with vertical and horizontal television synchronization signals, is colored by a color coding section 15, and a color pattern diagram of 15 colors is displayed on a color monitor 16. Ru. On the other hand, the image data quantized into 16 levels of gray data is transferred to the digital-to-analog converter 17.
The signal is converted into an analog signal and displayed on the black and white monitor 18 as a black and white video signal as shown in FIG. In addition to displaying the image data on a monitor television, a printer 19 and an X-Y plotter 20 are used as image output devices so that the image data can be kept at hand as materials.

FIG. 5 is a circuit diagram of the color coding section 15 for displaying the 4-bit image data read from the image memory 9 as color data of 15 colors.
The 4-bit image data for each picture element read from the image memory 9 is sent to the register 21.
The register 21 is composed of 4 bits, and the most significant data are R, G, B, and C, respectively.
R, G, B, and C are sent to buffer circuit _a1 ,
R, G, and B together with C constitute gate circuits b ₁ and b ₂ .
When C is logic 1, gate circuit b ₁ is opened;
R, G, and B drive analog switches d ₁ , d ₂ , and d ₃ . When R, G, and B are logic 1, the analog switches d ₁ , d ₂ , and d ₃ are conductive, and the 5V reference voltage is applied to the color cathode ray tube drive circuits e ₁ , e ₂ , and e through the buffer circuit c ₁ . ₃ , and a combination of the three primary colors of red, green, and blue is displayed on the color monitor 16.

Next, when data C is logic 0, the gate circuit
b ₂ is opened, R, G, B are analog switches d ₄ ,
Drive d ₅ and d ₆ . From 5V with resistors R ₁ and R ₂
The reference voltage reduced to 2.5V is applied to the buffer circuit c ₂ ,
When R, G, and B are at logic 1 via c ₃ and c ₄ , cathode ray tube drive circuits e ₁ , e ₂ , and e ₃ are driven via analog switches d ₄ , d ₅ , and d ₆ . By halving the reference voltage to 2.5V, intermediate colors of red, green, and blue can be displayed on the color monitor 16.

When R, G, and B are logic 0, buffer circuit b ₃
The analog switches d ₇ , d ₈ , d ₉ are driven through the circuits, and the inputs of the cathode ray tube drive circuits e ₁ , e ₂ , e ₃ are grounded and become logic 0. Moreover, when all three of R, G, and B are logic 0, the cathode ray tube drive circuits e ₁ , e ₂ ,
Since all inputs of e ₃ are logic 0, C has no relation to logic. In the above manner, R,
Fifteen colors can be displayed on the color monitor 16 from 4-bit image data of G, B, and C.

FIG. 6 is a flowchart of the image capture program. When a start pulse is sent to the analog-to-digital converter 8 in FIG. 2 during image capture, reading of the image data digitally converted into 4 bits is started. When high-speed capturing such as video is required, the image data is transferred to the RAM memory 13 via the input/output interface 12.
will be forwarded to. Further, in the case of a still image such as a photographic film, it is transferred to the floppy disk 14. Image memory reading, control of programs in the RAM area, floppy disk control programs, etc. are composed of subroutines written in machine language.

The image data stored in the image memory in this way is processed by a processing program. The flowchart is shown in FIG.

In FIG. 7, first, scales for the x and y axes are input in order to make the two-dimensional image data easier to see. Next, input the processing area designation and frame as parameters. Since the oil spreads continuously, smoothing and data correction processes are performed to remove singularities that occur during photography or data processing. In the case of the present invention, since the shading of the image is divided into 15 stages, there is little need to perform data averaging processing, so it is sufficient to perform only data correction processing. The data modification process is as follows. If the data of only one picture element is three or more steps farther apart from the data of nine surrounding picture elements at a point away from the oil diffusion region, the data of that picture element is replaced with the data of the surrounding picture elements.

The data processed in this way is sent to the floppy disk 14 and stored there. Next, as data processing for each stage, frequency distribution and area calculation processing is performed. Routine processing such as reading or writing data from the image memory uses subroutines written in machine language.

FIG. 8 shows a program for plotting the processing results on an X-Y plotter. First, the XY plotter is initialized, and then the drawing area is specified. Next, to determine the screen size, 1
By inputting the size per picture element as a parameter, reduction or enlargement of the screen is specified. By selecting a drawing pen, the pattern part, scale, and character part are color-coded, and then drawing is performed according to a drawing command. An example of drawing is shown in FIG.

When a microprocessor is used to perform arithmetic processing on an image, the processing time is generally long because the speed of the arithmetic processing device is slow. As a language used for processing programs, the BASIC language is easy to understand even for beginners, but because it is an interprinter method, the processing speed is slow, and it takes about 1 hour to read one screen of data.
It takes time. Machine language is superior in terms of processing speed, but it requires learning machine language instructions and it is not easy to change programs. However, in the present invention, BASIC language and machine language are used together, transfers with external devices and peripheral devices are performed in machine language, other processing is performed in BASIC language, and BASIC programs are converted into machine language by an integer type compiler. After converting it, we found that the processing time could be reduced by about 30 times compared to using the BASIC language. This compiler system allows data output processing to be performed at the maximum speed of the printer or XY plotter.

Next, a method for analyzing the oil diffusion situation in an oil spill experiment conducted using an experimental water tank will be explained.

FIG. 10 shows the temporal change in oil thickness measured using an oil thickness measuring device. At the mouth of the bay, approx.
It reaches a constant thickness in 50 seconds. At a point 2m away from the mouth of the bay, the thickness of the oil gradually increases as it spreads.
Approximately 300 seconds after the outflow, the thickness becomes almost constant.

When analyzing the oil diffusion situation, we first measured the shading of the photographic film obtained in the experiment.
Using the solid-state imaging television camera 7 in the figure,
The transmitted light of the photographic film 6 is measured. While watching the black and white monitor television 18, the upper and lower limits of the video level are adjusted using the analog-to-digital conversion setting volume in the analog-to-digital converter 8 so that the oil diffusion situation is displayed. The oil density pattern thus measured is displayed on the color monitor 16 in 15 colors.

FIG. 11 shows the relationship between the concentration of oil on the photographic film and the actually measured thickness of the oil. The relationship between the oil thickness d at a position 2 m from the bay mouth and the density level K of photographic film in the vicinity can be approximated by the following empirical formula: d = 0.144K + 0.16 (mm) (1). The arrows in the figure indicate an average distribution between the two concentration levels.
The amount of spilled oil can be calculated based on equation (1).

FIG. 12 shows the frequency distribution for each concentration level regarding the spread of oil at each time after the spill. At concentration level 0, the oil thickness is 0.16mm.
As the number of concentration levels increases, the oil thickness also increases, and at concentration level 14, the oil thickness is 2.2 mm.
becomes. Since the oil spreads in an almost semicircular shape, you can see that the thinner parts of the oil spread at the speed of the fish.

FIG. 13 compares the amount of oil Qth that actually spilled and the amount of oil Qcal determined by analysis.
The amount of spilled oil was calculated by multiplying the thickness of the oil by the area of one pixel. Since the amount of oil spilled from the spill source was measured during the experiment, this value was compared with the value determined by image processing. The oil amount Qcal determined by image processing is given by Qcal= ₁₅ 〓 ⁱ⁼¹ di·Si+q ₁ +q ₂ ...(2). Here, di is the thickness of the oil, and Si is given by Si=ΔS×Ni. △S is the area per pixel, Ni
is the number of density levels i. q ₁ is the amount of oil in the bay,
q ₂ is the amount of oil under the support rod of the oil thickness measuring device. q ₁ ,
As shown in FIG. 10, q ₂ was calculated from the output value of the oil thickness measuring device and the area at each time. The actual amount of oil spilled Qth,
The amount of oil determined by image processing, that is, the calculated amount of oil Qcal determined by equations (1) and (2), are almost the same until 200 seconds after the spill, but after that, the Qcal becomes smaller. The cause of this is
In image processing, oil thicknesses thinner than approximately 0.2 mm are ignored, but the area of this thin layer rapidly expands over time, causing errors in the amount of oil in this area.

FIG. 14 shows the relationship between the oil spill elapsed time and the oil spreading area. In oil spill experiments, the temporal change in the oil spreading area is an important parameter. Therefore, we extracted the areas with oil from the image data and calculated the area of spread. The empirical formula showing the relationship between elapsed time and spread area is obtained from the figure as Soct ^1.14 . Here, S is the spreading area and t is the elapsed time. Since the area S is the square dimension of the length L, Loct is ^0.57 . It has been revealed that this agrees well with the spread law of theoretical analysis in oil spill experiments, L∝t ^7/12 (7/12=0.58).

As explained above, according to the present invention, the density distribution of a photographic image of the oil diffusion situation is quantized into 15 density levels, and the thickness of the oil at specific points in the image is continuously measured. By correlating the concentration level near a specific point with the directly measured oil thickness, analysis is faster and simpler than conventional analysis methods, and the spread of oil over a wide area can be measured with high accuracy and reliability. It can be used in laboratories and production sites due to the minimum necessary functions, and is extremely effective in improving work efficiency and saving labor compared to conventional methods and devices.

[Brief explanation of drawings]

Fig. 1 is a three-dimensional diagram showing the configuration of an oil diffusion experiment to which the present invention is applied, Fig. 2 is a block diagram of the entire apparatus implementing the present invention, Fig. 3 is a video output signal of a solid-state imaging television camera, The figure shows the digital-to-analog converted video signal, Figure 5 is the color coding circuit diagram, and Figure 6 is the flowchart of image capture.
Figure 7 is a flowchart of the image processing program.
Figure 8 is a flowchart of the drawing program, Figure 9
The figure is an example of drawing, Figure 10 is the measurement result of the oil thickness measuring device, Figure 11 is the relationship between oil thickness and concentration level, Figure 12 is the frequency distribution of each concentration level, Figure 13 is
The figure shows the relationship between the amount of oil actually spilled and the amount of oil determined by analysis, and FIG. 14 shows the relationship between the elapsed time after oil diffusion and the spread area. 1: Oil diffusion experiment tank, 3: Photographic camera, 4: Oil thickness measuring device, 7: Solid-state imaging television camera,
8: analog-digital converter, 9: image memory, 15: color coder, 16: color monitor, 20: X-Y plotter.

Claims (1)

[Claims] 1. The spread of oil and pollutants spilled into the ocean, rivers, lakes, etc. and the combustion area in furnaces and combustion chambers are photographed at any time using a camera, and on the other hand, a direct measurement method is used to identify specific points in the grayscale image. After continuously measuring physical quantities such as thickness, density, and length, and photoelectrically converting the grayscale image taken by the camera, the analog-to-digital converter quantizes it as density level data and calculates the density level for each pixel. is stored in a storage medium, a density distribution map is drawn with an X-Y plotter, and the physical quantity for the image density is determined by correlating the density level near the specific point in the image with the physical quantity measured by the direct measurement method. 1. A digital image processing method characterized by determining the distribution of physical quantities in the image and measuring the distribution of physical quantities in the image. 2. A camera that photographs contaminated areas, combustion areas, etc., a photoelectric conversion unit that converts the photographed images into electrical signals, an analog-to-digital converter that quantizes the output of this photoelectric conversion unit as concentration level data, and a storage medium for storing output data;
A measuring device that directly measures physical quantities such as concentration and thickness at a specific point in a photographed image; and a comparison detection unit that compares the output value of this measuring device with output data near the specific point of the photoelectric conversion unit. , and an X-Y plotter for storing density patterns.