JPS603060A - Collecting device for image contour - Google Patents

Abstract

PURPOSE:To extract the contour of a body easily and continuously by generating a combination of signals of four picture elements which contact a contour, and converting them into quinary signals indicating a tracing direction and storing in a picture memory. CONSTITUTION:The output of an ITV camera 72 is converted by a comparator 76 into a binary-coded digital signal, which is supplied to a TV73 for monitor display; and one horizontal-scanning-line signal delay part 77 and one picture-element-signal delay part 78 obtain four adjacent picture-element signals, which are inputted to an output conversion part 79. The output conversion part 79 outputs a quinary-coded signal indicating the tracing direction of the picture contour. This quinary-coded signal is stored in the video memory 81 to form a picture whose edge can be traced.

Description

DETAILED DESCRIPTION OF THE INVENTION An object is captured by an optical 1i conversion device such as a camera, and the grayscale image is converted into a black and white 21 image, and then the 1i product is obtained.

This invention relates to a device for determining the quality of an object's shape and identifying its surroundings. The principle of this device is I
i' The boundary between the object and the background captured on the screen captured by a V camera, that is, the outline of the object, is plotted using the in-screen lip standard.
dI calculations are performed on the area, perimeter, etc., and pass/fail judgment and identification are performed based on these. In Juto's method, the image processing unit uses a raster scanning method in which the scanning line 100 is horizontally scanned sequentially from left to right from the edge to the bottom of the screen, as shown in FIG. Most likely it's a V camera.

In line with this, collection of layered data on contour points was performed according to raster scanning. For this reason.

The results of the contour line take are arranged in 16 series of Lascou scans - (+j+i planes are scanned), so the contour line I!
The V mark is obtained, but the drawback is that the 9p mark is G1.
? Molecule for each object to be l4lI. There are 12 points that must be considered when it comes to ip L/. The first (χ(i) general image contour acquisition device will be described as a roughly flora block.

21 is an illusion, 22 is an image! ITV camera that captures lj2+ with a line, 23 is a monitor TV, 24 is a synchronization signal generator, 25 is the same J9J '70 Tsuk generation R] (
, 26 is an etch extraction unit having a function of extracting contour points of the object 2J, and 27 is a contour point coordinate memory.

Figure 2 shows the scanning line 1 of 23 monitor television art dealers.
00 and the image 30 of the object 21 are indicated as 1 to 16, respectively, indicating that the contour v1 and the marker number are discovered in order.
...X11i + )'+6 [! ] Six markers are stored. As is clear from Figure 2, initially 1, 2
Sections 3 and 4, and sections 5 and 6 are each judged to be three independent objects. However, as the judgment progresses, 1, 2, 8, 4, 5, 6, 7, 8°9, 10.11.
, 1.2. IJ 1.4 is 5□ 6゜1 in one object
1, 12, 15, 16 or one object. In order to determine stiffness in this way, there is a drawback that a complicated determination must be made using coordinates in the vicinity of contour points.

Tree yJ and brightness were made to reduce the compound interest judgment described in -1, and when an object is moved horizontally with a photoelectric conversion device i"1"
Take the f-line and subdivide the 1-light image into pixels and 2fli
The conoparator that converts to tenotaru and the output of the comparator delays the time equivalent to one horizontal scanning line.
Water 1' - scanning line borrowing delay means and horizontal scanning line 11i
A 1-pixel delay-F stage that delays a time corresponding to llll elements, and the output 1. of the conopalator. ) Oyohi 1) 1
Pixel LT ``j Signals ゛C and IJ delayed by delay means
The signal A delayed by the scanning signal delaying means and the signal I3 delayed by the one-pixel signal delaying means - C are simultaneously manually inputted into the image contour line jIA in the inward direction ■ 5
and an output W converting means for converting 7'3 into a preformed signal.
The image contour line collecting device 7''' which uses the obtained data as 14 features creates a combination of 4 pixels that are in contact with the 7 contour lines, and outputs these 4 signals to the same 114 output converter and converts them into contour points. The purpose of this is to easily determine the connection of contour points by dividing the five-value code indicating the tracking direction of the contour points.

DESCRIPTION OF THE PREFERRED EMBODIMENTS The image contour collecting device according to the present invention will be described in detail with reference to the drawings.

In Fig. 3, an object 51 is reflected in the vertical In point (image 1iii 50 consisting of 111,111 points, and the object 51 is
The dorsal shape of the object 51 is shown as having a binarized level of "0" and an "i" arch.Explain the J method to create an image in which only the outline of the object 51 is extracted from such an image. , whether a certain 1, y constant point (i, j) is a point or a contour line (hereinafter abbreviated as edge) is determined from other three adjacent pixels (i-1, j-1).
), (i-1,, i), and (i, j-1). In other words, if you cut out 2 pixels vertically x 4 pixels horizontally, each pixel can function in two ways, ``Oj'' or ``1''.
Therefore, as shown in the binarized image 411 (2) at the left end of FIG. 4, only 16xil combinations are possible, and only by combining these, it is possible to trace the outline of the object. For example, the third
60 in the figure shows the state where the object 51 does not exist, 61 shown in FIG. . Among these, all pixels other than 4 pixels are completely "0" or completely [-1] indicate that the corresponding point is etched. r (i, , i)
i, i-0 table 4 and 4 pixel values ai, i, ai
-1,j+ll1-1. , i - 1 + 111, , i I performed 2, 4, and 8.1 checks on l, respectively, and added them together, and the value is 16] ml -1 for each m silk combination.
: can be assigned. These (, indicates that there are 16 types of silk combinations notified in t22 in decimal notation.-1
The weight of 111 is calculated by the following formula.Explaining an example in Figure 5, +20 is the pixel value i+'r.
The nail polisher 53 has an outline 7X1 (
.

] 20 and 53, respectively, are multiplied by the corresponding values.
When it is 1, it becomes f-like of 1. In addition, ] 20 nails 1 kei
, t This is for converting into a hexadecimal code, and the positions of each number of 1°2.4.8 can be set to other combinations.

In this way, l21.2.・+1 H, i == 1
+2,...m, if you do -CE'f MINI J KEGO 1 parent for ``Sutete'', 16 types of coats will be created over the entire screen.

Next, carve each of the 16 types and perform the conversion to make the ecchi 16/+ (J. In other words, convert the 16 types of code 10,1 with 16 chains to the 5 types of code 1'c! Convert to i, i.

This conversion is to indicate the direction of etch tracing, as shown in Figure 4, for example, if you look at the object to the right, you can trace the etch to the right.
Down, no,, ru4! T! Figure 4 summarizes the results of code conversion for five types, one type without tracking direction, and one type without tracking direction.

In this way, the meseri device i! creates a quinarized image code from a binary image. If you store it in f, -carrier "0"
If you memorize the pixels that are not included, that is, the etch points, and open the hospital,
The contents of the memory indicate the direction of the next etch point and can be extracted in a formula that draws the outline of the object as a single closed curve.

Next, the method according to the present invention is shown in FIGS. 6 to 11.
This will be explained using a specific example. In this example, an object 51 of 4 pixels in the vertical direction and 6 pixels in the horizontal direction is manually processed and
In the 1st image 110, I have 5 vertical pixels and 7 pixels, which is 0. This is because, in principle, in the 4-n dynamic method (according to the invention), each pixel is inflated by 1 pixel.

When I (j!! 1f1150 H object 5 on the right and left) in Fig. 6 is scanned by the scanning line I418, the object 51
The pixel where exists is as shown in Figure 7 (numerical value [1j-b';
The pixels in the portion where the object 51 is not provided are not given the numerical value "0" and the remaining pixels are binarized. Next, as shown in Figure 8 (the "1" part of the binarized pixels is 61 in Figure 4).
Since it corresponds to 411 in the binarized image shown in FIG. Next, in the pixels of the etched part in FIG. 9, the four pixels indicated by 54 are as shown in FIG.
2", ζ2('s 4 pixels indicated by 55 are "6" and (
The hexadecimal code is converted to 11'v, and the edge and bottom edge parts are ``4'' or one extra pixel due to the 4 pixels of 56 according to the tree/-j method, ':lI 1tlii
Therefore, one extra pixel worth of hexadecimal code is added to the bottom left corner.
However, this method does not have any secondary functions.Next, as shown in Figure 10, etch 1XIX 1
]11 code is converted into CII 5-value code>'+ +
As shown by the arrow 110 at + (1t, -) (approaching the contour part as one closed loop), it is possible to do 1j
Becomes Ding Noh. In addition, Figures 10 and 11 show Elan 1.
A pixel to which 1-+5J other than Xi (1-+5J) is displayed is displayed as "0". Even if "15" is left as is, it will not become 1f.

The image contour collecting device according to the present invention will be described in addition to the method described in θ(ni). In FIG.
i conversion unit, 73 is a TV for monitor display, 74 is I TV
Water 517 phantom to camera 72 - synchronization, vertical synchronization J' JJ belief - J '+l E1 閤175
is a -7 star crotter generator for synchronizing the entire device.

76 is a comparator that converts the analog image signal to 10".

The 211iT knuckle signal of 1"+J is converted.

Reference numeral 77 denotes a 1-scanning line delay unit for delaying the time corresponding to one horizontal scanning line of a television picture dealer.

As a result, the output of the 77 is compared to human power, and the J Lasque of the TV image direction is 635 μS (microseconds), which is 6 B
, 5 // 8ill.

78 is the same (, -7 seconds, I!f extension i'91, but the screen water 'I' direction is divided into 1111111 elements 1141
I jiij element I 11 that delays the time equivalent to pixels
The bow delay section provides a delay of 6352 n/ls.

79 is the output conversion ff1(, and the 16-value code 1~(bji
) is converted into (C!i, i ) by 5Mi & Co1, and the 4-tree signal power of A, B, C, and D output from 76, 77, and 78 is input into the address of the zero output converter 79. ,
Outputs the memory contents indicated by that address. /′R are signal delay units 77, 78 . The function of the output conversion memory 79 will be explained in detail.

The binarized image that is the output of 76 is instantaneous according to raster scanning, and at the time of In
Output the point take. In the method according to the present invention, it is necessary to take this particular point and its 45'7 position to the right of the door, and to the 10th right of the door.
As shown in Figure 3, there are two scanning lines 1. It is necessary to take out the four pixels 54 scanned by OO at the same time. 1 horizontal scanning line signal delay 7゛τlX77 corresponds to 256 horizontal pixels on both sides.
If this is the case, it will consist of 256 stages of Nftrenostar or Crow delay line, and will take one scanning line time, that is, L2 shown in FIG. 14! 1] Extend the raw signal by 1JX.

The 11II11 element signal delay unit 78 is a J) type flip-flop and has a delay unit for one pixel, and has 256 pixels in the horizontal direction on both sides.
If this is done, the signal will be delayed by 2/256=tl. For the output of the two lumberrators 7G, A is delayed by 1,2°1(is 1.2 + (, l213.(delayed; The output conversion unit 79 manually inputs data in 11p columns.Therefore, there is raster scanning]
When you come to i,,i),
) point +”+j+C has (i-1,,i
) point 21 straight take a1-1,1. To H! , :1.
The two-fold theta ai,,il+A of the point (i,,i-1)
is a 2f direct take l1l− of point (i −1,,i−1)
You can get 1+1-1 respectively. ] In Figure i+4, as an example, the scanning lines and images of the 2nd tree 1, 1 and 3rd tree; number 101.
Clock signal 102° Signal lo3 of pixel 54 in FIG.
．． 104 relationships are shown.

These four signals A, IJ, and C were generated manually at the same time.

D is the output exchange part 79, 4hi, 1.16 as shown below.
l+i11 is applied to the hexadecimal data.

``i, , i -1(B), 20 hits -1n 1., H(1+) with 21, 1 = 2a +],,''(C), 22 hits
1-=4”i 1.j-1(A), 28
L, l--8, i.e., the aforementioned l Ill J-I
Xa i, J-]-1-2X+i i, , 1-1-
4Xn; 1. J + 8 This 4-hit hexadecimal code from "0" to "15" is then converted into a 5-value code (Oi, i). This converter consists of It and OM, and as shown in Figure 4, b i
, i and Ci , do the correspondence between i and i to create a five-value code c
The output of ij is stored in image memory 81 to create an image that can be traced by etch.

Reference numeral 80 is for the image memory 81; an address generator including IF takes m x and n in Fig. 3 in chronological order jjj.
This is for storing in the ij image memo 1J81. In addition.

The content of the mxn take at this time is the 5-value data for etch tracking shown in FIG. 83 is 2 shown in FIG.
Contour 1 marks are collected and stored in a memory similar to that of 7.

Once the etch tracking image of 1+r+i' is created in the image memory 81, the computer 82 starts collecting 0 edge marks. This ticks the edge of the 81 screen and turns up the part 7, which is not 10'', of each u11 screen. - If the work is selected, the work, date] -
--The 5-value code ((,:',i,i) indicates the Jj direction of the next etch, making it easy to see which part of the memory should be accessed.Even the address at the time of discovery can be found. If this happens, the contour line will be IJ,'il, and the tracking will end at this point.

In the image contour line collection device of the present invention described above, a combination of four pixel signals touching one contour line is created and these signals are simultaneously inputted to the output conversion section and quinarized to indicate the tracking direction. If the image is converted and stored in the image memory, the outline of the object can be extracted continuously (in an imozuru style) for each object.

[Brief explanation of drawings]

Fig. 1 is a block diagram of a general image contour collecting device 1'7+'', εi% Fig. 2 is a block diagram of the monitor television 23 shown in Fig. 1.
01 example explanatory diagram, Fig. 3 is an explanatory diagram of a screen subdivided into III x It and a binarized image of an object, and Fig. 4 is a combination of a binarized image and a 5-valued image at 4 pixels. An explanation of the coat, and Figure 5 is an example illustration of the coat.
6 to 11 are a series of explanatory diagrams of coat conversion and contour tracing of each pixel of an object. 7. FIG. 12 is a block diagram of an image contour collecting device ijj according to the present invention. Explanation of the relationship between and the pixels of the object> 1. Figure 14 is the glue in Figure 13-
(I-inches 1 to 1. - Synchronous signal generator, 25... Synchronous black 2, ri generation part, 26 - Etch drawing 110! ((,27
Contour point 1 Tokyo marker memory, 80...image, 50...screen,
51...object, 52.53°54, 55.56.5
7... 4 1+lii elements of contour line 1 (2nd age image),
60...4 pixels that are not present in the object. 61... Four pixels existing inside the object, 71 Object, 72... Photoelectric conversion unit 273 Telephone for monitor display. 74... Synchronous signal generation 7H + 75 - Master black signal generation section, 76... Comparator 1 notator, 77... Signal delay section equivalent to 1 horizontal scanning line time, 78. Delay 81 (, 79... Output conversion memory, 8
0...Write address generation jIX, 8]...Image memory, 82...Co, bitter, 83...Contour p(...Memory for target acquisition storage, +00...Scanning line, 101...Image Believe,
102. Kuro, Ri (('1 bow, 103. 104.
jlij element signal, ]10...outline part pixel, 120...
Weight Figure 4/20

Claims (1)

[Claims] A comparator captures an object using 3F scanning lines using a photoelectric conversion device r1, divides the grayscale image into 111 pixels, and converts it into a binarized tental signal, and 1 delays the output of the comparator by a time corresponding to 312 scanning lines. Water 3 scanning line signal delay means, 1 pixel delay/delay means for delaying the time corresponding to 1+1lii element of water "1' scanning line, and comparator outputs I] and I) are delayed by l1ijii gray signal delay means. The signals C and U-D are delayed by one horizontal scanning signal delay means, and the signals A and a< A are delayed by one pixel signal delay means. 1. An image contour collecting device comprising: an output converting means for converting into a quinarized signal indicating a tracking direction.