JPS5826521B2 - Graphic analysis method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method and apparatus for measuring the area, length, etc. of an image or figure.

There are various methods for analyzing images obtained using a microscope, etc., but methods that quantitatively determine the area and length of many microscopic images or figures within a fixed field of view are particularly useful in shortening the measurement time. Automation is being attempted to achieve this goal.

In general, automatic analysis of images (figures) scans a fixed range of visual field with a series of horizontal sweeps, and divides each horizontal sweep at fixed time intervals to divide the entire visual field into a large number of pixels divided into squares. It is disassembled and processed.

Furthermore, video signals are often distinguished into two types of signals depending on whether each pixel is above or below a certain threshold, or whether it is between two thresholds. .

Furthermore, in conventional methods, (1) a method of storing all the information of each picture element within one field of view and then reanalyzing each individual figure; (2) a method of scanning picture elements and sampling their video signals; A method for sequentially analyzing shapes each time
(3) A method has been used in which information regarding figures is temporarily compressed and stored and later analyzed for each individual figure.

According to the first method (1), a storage means with a very large capacity is required, and according to the second method, the analysis processing performed at each sampling time is complicated, and therefore it takes a long time to analyze the entire field of view. It was normal to waste time.

The present invention, which will be explained below, belongs to the third method, and is based on the premise that when the visual field is divided into a large number of minute picture elements, the video signal is discriminated into binary values of O or 1, and the storage The feature is that the capacity of the means can be reduced and the processing time can also be shortened.

Now, when the image in the field of view is quantized to black (1) or white (0) as shown in Figure 1, let's first calculate the area, length, etc. of each black figure as shown in Figure 2. Only one figure shown will be explained.

First, focusing on the change between the information on the horizontal sweep line currently being scanned and the sampling information on the previous line, consider the flag signal F that indicates the global nature of the figure.

Examples of F include Fig. 3 a, b, and c. Fl, F2, etc. according to the figures shown in d and e. F3°F4. Five types of F5 signals are provided.

Fl is set when the information on the current line is all 1 and the information on the previous line is all 0, that is, when the current line reaches a convex portion above each figure.

F2 is the opposite of Fl, and is set when the information on the current line is all O and the information on the previous line is all 1, that is, when the current line reaches the downward convex portion of each figure.

F3 is set when both the current line information and the previous line information are all followed by l, that is, inside each figure.

F4 is set when the information on the current line is all 1 and the information on the previous line is 1 or 0, that is, when there is a 0 in the middle of l.

F5 is set when the information on the current line becomes 1 or 0 and the information on the previous line is all 1.

Note that no particular flag is provided for the hole in the figure.

This is as shown in FIG. F5, two F3s, F
This is because the flags are set in the order of 4°F3.

These flag signals make it possible to collectively process information obtained by scanning a large number of figures within one field of view for each figure.

Next, the length information N of the portion where each figure intersects with the currently scanned horizontal sweep line is counted in units of picture elements.

The area ΣN(N
The vertical projected length of each figure is determined from the flag signal F and the number of horizontal sweep lines that intersect with the figure.

For example, in the figure of FIG. 2, F and N are as shown in Table 1.

Since Fl stands on the lth line, it is determined that a new figure has started.

Since F3 stands from the l+1th line to the 1+4th line, it is determined that the figure that started on the lth line continues downward.

On the l+5th line, F2 is standing, so the figure is 1+4.
It is determined that the process has ended at the line.

And the area of the figure in Figure 2 is ΣN=2+5+7+5+2
+0=21 picture elements, and the vertical projection length is 5 picture elements.

However, in actual image analysis, figures of various shapes must be handled.

Since a plurality of depressions, that is, depressions, may occur on the same horizontal sweep line, information on the number of depressions (concavities) D1 and D2 is added to F4 and F5.

In the figure of FIG. 4, there is one depression each at F4 and F5, so F3. F5 (D2=1), two F3, F
The flags are set in the order of 4 (DI=1) and F3.

For example, in the case of the figure shown in FIG. N.D.I. , D2 are as shown in Table 2.

Since Fl stands on the lth line, it is determined that a new figure has started.

On the l+1th line, F5 is standing and D2-1, so it can be determined that the figure has been divided into two, and although the figure has been divided into two, they intersect with the horizontal sweep line for a total of 2 pixels. It turns out that it is.

Since there are two F3s on the l+2 line, it is determined that the figure divided into two on the 1+1 line continues below.

On the l+3th line, F4 is set and DI=1, so it can be determined that the figures separated on the l+1th line are reunited on the 1+3rd line.

Since F2 is set on the l+4th line, it is determined that the figure is finished on the 1+3rd line.

And the area of the figure is ΣN=4+2+1+2+4+0=1
3 pixels, and the vertical projection length is 4 pixels.

Further, when F4 and F5 occur on the same horizontal sweep line as shown in FIG. 6, F4 is processed with priority over F5.

Since Fl stands on the lth line, it is determined that a new figure has started.

Since Fl and F3 are standing on the l+1th line, it is determined that one new figure has appeared on the left side of the image (the position swept forward) from the figure that started on the lth line.

Then, it is determined that the figure that started on the l-th line continues directly to the bottom of the construction.

On the l+2 line, F4 is standing and Dl-1, so the figure that was divided into two until the l+1 line becomes 1+2.
You can tell by looking at the line that there was a rush.

Furthermore, since F5 is also standing, it can be determined that the combined figure is further divided into two parts (from D2=1).

This is why I am giving priority to F4 over F5.

On the l+3rd line, F2 stands first, so 1+2
It can be determined that the left part of the figure divided into two at the line has been completed, and since F3 is next, it can be determined that the right part of the divided figure continues.

Since F2 is set on the l+4th line, it can be determined that the figure is finished.

The area of the figure in Figure 6 is ΣN=1+2+3+6+O+1+
0=13 picture elements, the vertical projection length is 4 picture elements.

No matter how many figures there are on the image, if F, N, DI, and D2 are written without changing the order of the figures appearing on each horizontal sweep line, the area, length, etc. of each figure can be measured correctly.

As mentioned above, the five types of flags as shown in Fig. 3, the length of the part where each figure intersects with the horizontal sweep line, the information on the number of depressions related to flags 4 and 5, and the information that flags 4 and 5 are on the same horizontal line. The area, length, etc. of the figure are determined based on the judgment criterion that flag 4 is given priority when set by the sweep line.

Next, a method and apparatus for specifically carrying out the above-mentioned measuring method will be explained.

Figure 7 shows the flag (F'l, the length N of the part where each figure intersects with the horizontal sweep line, and the number D of depressions related to F4 and F5.
1. FIG. 2 is a block diagram of means 15 for generating D2.

A video signal applied from an input terminal X is first quantized into binary values of l or 0 by a quantization circuit 1, and then applied to a shift register 2.

The shift register 2 has a capacity equal to the number of picture elements divided on the visual field scanning line (for example, 500), and when new sampling signals are sequentially applied from the left, each storage position moves to the right. When the right end is reached, the output signal is applied to the left end of the second shift register 3.

If the sampling positions of the two stored values at the left end of these shift registers 2 and 3, that is, the four signals stored at the atbtctd position, correspond to the field of view to be analyzed, each a is shown in FIG. ', b', c', d'.

In this way, if the quantized video signals of the four picture elements adjacent to each other are simultaneously detected and the entire field of view is substantially scanned as a unit of these four picture elements, the arithmetic circuit 4 calculates the N value as described above. can be measured.

Further, the global nature of the figure is identified by a flag generating circuit 5 attached to the arithmetic circuit 4.

The number of depressions is determined by the calculation circuit 6 attached to the flag generation circuit 5.
It is counted by

N value, flag, DI value, D found in circuits 4, 5, and 6
The binary values are temporarily stored in the register 7 or output directly.

When register 7 is used, the memory signal is updated at the right end of each figure within the field of view.

Information F from means 15. Depending on when to process N, DI, and D2, several methods are possible, as described below.

The first method is the information F, N, DI from means 15,
This is the 9th one that tries to store D2 as is in memory.
The configuration is as shown in the figure.

Reference numeral 8 in FIG. 9 denotes a memory having a function of storing the information sent from the means 15 separately for each horizontal sweep line, and its output is a l frame end signal (l visual field scanning completion signal) or a wind end signal. Gate circuit 12 opened by (signal emitted at the lower boundary of the measurement area of the visual field)
The signal is applied to the analysis calculation circuit 9 via.

The analysis calculation circuit 9 performs corrections such as grouping identical figures together using the flag signal F and the number of depressions D1D2, and calculates the area, length, etc. of each figure.

The output is stored in memory 10 or sequentially output device 11
It is output from

The second method is by means 15 of the information F and N.

DI and D2 are temporarily stored in memory, and the figure is analyzed at the end of each line scan.
The configuration is as shown in Figure 0.

13 in FIG. 10 indicates information F and N for one horizontal sweep line.

A gate circuit 14 whose output is opened by a line end signal (one horizontal sweep line number) or a wind end signal (a signal emitted at the right edge boundary of the measurement area of the visual field), which has the function of storing DI and D2. The signal is applied to the analysis calculation circuit 9 via.

The analysis calculation circuit 9 receives the flag signal F and the number of depressions Di, D2.
The area, length, etc. are determined from the measured figures by sequentially performing corrections such as grouping the same figures together.

The output is stored in the memory 10 or sequentially output to the output device 1.
Output from 1.

According to the present invention, as described above, it is possible to perform image analysis in a short time using a processing device that is relatively smaller than the conventional device.

[Brief explanation of drawings]

FIG. 1 is a schematic diagram showing the distribution of figures analyzed according to the present invention, FIGS. 2 to 6 are schematic diagrams showing the principle of image analysis according to the present invention, and FIG. 7 is a schematic diagram showing the main parts of the apparatus of the present invention. FIG. 8 is a schematic diagram for explaining the operation of the device shown in FIG. 7, and FIGS. 9 and 10 are schematic diagrams showing embodiment devices of the present invention. 1...-quantization circuit, 2,3...-shift register 4...-arithmetic circuit, 5--...-
Flag generation circuit, 6...- Arithmetic circuit, 7...
-Register, 8, 10, 13... -Memory, 9.
...Analysis calculation circuit, 12,%14...-
Gate circuit, 11... Output device.

Claims (1)

[Scope of Claims] 1. A measurement field of view is divided into a large number of minute picture elements, and a series of 9 horizontal sweeps is performed to distinguish the video signals obtained from the picture elements into two types of signals, and the picture elements having the same type of video signal are In a method of analyzing a figure formed by a set area of
When we measure the length information N of the part where each horizontal sweep line intersects with each figure, we can measure the length information N of the part where each horizontal sweep line intersects with each figure. If the information on the line is all 1 and the information on the previous line is all O, then the information on the current line is all O and the information on the previous line is all 1, then the information on the current line is 1. If the previous information is all l, and the information on the current line is all l and the previous information becomes 1 or O, then the information on the current line becomes 1 or 0. A flag signal corresponding to each case is obtained depending on the case where all the information on the previous line is l, and the length information N is determined for each figure using the flag signal indicating the global property of the figure. A figure analysis method characterized by measuring the length or area of each figure by performing arithmetic processing as a set. 2. Means for scanning the field of view to be measured by a series of horizontal sweeps and dividing the field of view into a large number of very small picture elements by dividing each horizontal sweep in fixed units; means for discriminating into different types of signals; - storage means for the video signal having a capacity greater than the number of picture elements included in a horizontal sweep; and means for reading out a video signal possessed by any picture element in two adjacent horizontal sweeps from the storage means; , if the information on the current line is all 1 and the information on the previous line is all 0 for a figure based on the information from the reading means, then the information on the current line is all 0 and the information on the previous line is If all the information is l, if all the information in the current line is l, the information in the previous line is all l, if all the information in the current line is l, and the previous information becomes 1 or O , the information of the current line becomes l or O, and the information of the previous line is all l, the flag signal corresponding to each case and the length signal N in each sweep direction are generated. A figure analysis apparatus characterized by comprising: means for generating the flag signal and means for determining the length or area of each of the figures based on the flag signal and the length signal N.