JPS62120185A - Image recording method - Google Patents

Abstract

PURPOSE:To largely reduce a required memory capacity and make a constitution of a system compact by recording information quantized by the number of bits smaller than a color picture on a black and white picture. CONSTITUTION:An input picture to be recorded is stored temporarily in a frame memory 3 and thereafter, divided into a part having the picture and a part having no picture on a screen. On the part having the picture, a control signal representing the position and the size is formed. Whether the input picture is the color picture or the black and white picture is discriminated by presence or absence of a color burst signal of a manual designation, and when it is the black and white picture, the picture information quantized by the number of bits smaller than the color picture is recorded. In the case of the color picture, the input terminal B of a selector 10 is selected, a quantized value of 8 bits read from the memory 3 is recorded in a register 17 and recorded on a disk memory 14 through the selector 10.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an image recording method, and particularly to an image recording method in which a color still image and a monochrome still image are recorded in the same file.

(Prior art) Normally, when recording television images in digital form, the sampling frequency is 10.73 MHz (equal to three times the color subcarrier frequency fsc) or 14.31 MHz.
It is recorded at MHz (4f, c) with a quantization bit count of 8 bits. Therefore, recording one screen requires 26 B or 350 kB, and recording a large number of still images requires a memory with an enormous capacity. Also,
Conventional recording methods require a memory for recording color images,
It is provided separately from the memory for recording black and white images.

(Problems to be Solved by the Invention) As described above, the conventional image recording method requires a memory having an extremely large capacity, and has the disadvantage that the overall system configuration is extremely complicated and expensive.

Furthermore, since the color image and the monochrome image are recorded in separate memories, there is a drawback that the configuration is further complicated. On the other hand, when considering broadcast images, the still images that are actually recorded often include text and cut pictures for supermarkets, and black and white images often exist only in a portion of the screen. In conventional image recording methods, information for an entire screen is recorded even for such an image, and the required memory capacity increases accordingly.

The purpose of the present invention is to solve the above-mentioned problems of the prior art,
It is an object of the present invention to provide an image recording method that can significantly reduce the required memory capacity and can therefore be implemented with a simple and inexpensive configuration.

(Means for solving the problem) The image recording method of the present invention divides the screen of an input image to be recorded into a part with an image and a part without an image, and for the part with an image, It records control signals representing the position and dimensions in this area and image information within this part, and further determines whether the input image is a color image or a monochrome image. It is characterized by recording image information quantized with a smaller number of bits.

(Function) In the present invention, after an input image to be recorded is temporarily stored in a frame memory, it is divided into a portion with the image and a portion without the image on the screen. A control signal representing the position and size of a certain part of the image is created. On the other hand, whether the input image is a color image or a monochrome image is determined based on the presence or absence of a color burst signal or by a manual instruction. For supermarket images where the input image is binary (white and black), a satisfactory image can be reproduced with 2-bit quantization, even considering the smoothness of diagonal lines, and even for black and white analog images, it is possible to reproduce a satisfactory image with 4 bits. It has been confirmed that bit quantization (16 brightness levels) is almost sufficient, and there is no need for 8-bit quantization as in color images. Therefore, for monochrome images, first 2-bit quantization or 4-bit quantization is performed.
By using bit quantization, the memory capacity can be reduced to 1/4 or 1/2.

In addition, images for supermarkets are generally character strings or cut pictures, and images are rarely present over the entire screen, but are unevenly distributed in a portion of the screen. Accordingly, in the present invention, image compression is performed so that information on portions that do not include images is not recorded, and information quantized in 2 or 4 bits is recorded only on portions that include images. In this case, a complete image can be reproduced by recording control signals representing the position and size of the portion containing the image on the screen. By recording only a portion of the screen in this way, memory capacity can be dramatically reduced.

That is, compared to the case where 8-bit quantized values are recorded over the entire screen, it has been confirmed that 2-bit quantization results in only about 4%, and 4-bit quantization results in only about 8%. Since the part of the image on a normal super screen is extremely small, when compared with the same memory capacity, it is 25 to 5 times faster than the conventional method.
This means that approximately 0 times as many screens can be recorded.

(Embodiment) FIG. 1 is a block diagram showing the overall system configuration of an image recording and reproducing apparatus that implements an image recording method according to the present invention. An analog image signal of a still image to be recorded, which is supplied to an input terminal 1, is converted into a digital signal by an A/D converter 2. This conversion is performed by quantizing, for example, 8 bits per pixel. The digital image signal quantized in this way is stored in the frame memory 3. In order to control the reading of this frame memory 3 by X and Y address generators 4 and 5 each having a presettable counter, input terminals 6 of these X and Y address generators are provided.
and 7 from the CPU 8. As a result, an arbitrary position can be specified, and therefore pixels at an arbitrary position on the frame memory 3 can be selectively read out.

First, the operation of dividing the screen into a part without an image and a part with an image will be explained. First, a signal is sent from the CPU 8 to the X and Y address generators 4 and 5, and the frame memory 3
The first line (Y=O) is selected, and one line of image signals is sequentially read out from the origin (XO, yo) in FIG. 2 while changing the X address. - When the reading of the 2nd line is completed, set Y=1 and read the 2nd line. The OR gate 9 calculates the logical sum of the upper 2 bits or 4 bits of the 8-bit image signal read from the frame memory 3, and the CP
Supply to U8. As shown in FIG. 2, when a non-black point is read for the first time at Y=Y, the output of the OR gate 9 becomes "1". This one point is where black becomes white, so it is a point on the outline of the image. In order to find the coordinates of a rectangle surrounding the image starting from this one point, next, as shown in Figure 3, centering on one point (the pixel indicated by P in Figure 3), eight adjacent The pixels are read out sequentially in a counterclockwise direction starting from the pixel immediately above. In this case, unless one point is an independent point, a non-black point will be read somewhere. In the example shown in FIG. 3, the OR gate 9 is
The output from is "1", and the fact that this point is not black is c
It is determined by rt, +s. Next, similar reading is performed centering on the newly found pixel P2, and pixel P3 is found. If such processing is continued sequentially, points p, , p2 . p, . . . PM are obtained, and the pixel returns to pixel P1 again. At this point, this operation ends. Next, the coordinates (XI, yl
), (X2.y2)・”(XN4*) on the X-axis and Y-axis, we can draw a rectangle surrounded by these minimum and maximum values (rectangle ABCD in Figure 2).
) will exactly contain the detected image. However, the figures are not necessarily unevenly distributed in a row; for example, when considering the character ``小'', it can be seen that three parts exist adjacent to each other. That is,
Generally, there are figures that are close to the figure shown in FIG. 2, but are far away from it. In such a case, if these figures are processed as separate parts, the number of parts containing images becomes too large, making signal processing troublesome. Therefore, in this example, the rectangle ABC obtained as described above
A rectangle PQR3 that is larger by W in both length and width is set outside D. This dimension W can be, for example, the height of a standard size character or the 172nd height thereof. Next, it is checked again whether there is a point where the pixels are not black in the corridor between these rectangles ABCD and PQR3. Using the point H' found in this way as the starting point, the coordinate string of the contour line (XI
' , Vt '>+ (×2 '*y2')...
Find (xn', yN'). Next, this coordinate string and the coordinate string (X, , yl
), (X2゜Y2)...(XN, yx), find the minimum value Min (x+, yl) and maximum value MaX (Xt + yt) on the X and Y axes, and create a new enlarged rectangle. Set ABC'D'. Next, a rectangle PQR' that is larger only in the X direction than the rectangle PQR3 described above by W
Set S', rectangle ABC'D' and rectangle PQR'
Check whether there is a non-black point in the corridor between S'. This process is repeated until no non-black points are found in the corridor. In the example shown in Fig. 2, no non-black point is found in the corridor between rectangle ABC'D' and rectangle PQR'S', so rectangle ABC
'D' can be detected as a rectangle surrounding a group of figures.

The CPU 8 calculates the upper left coordinates (X
I, yl) Width △XI=Mag (XI) Lh(XI
) and height △Y, = MaX (Vt) Mt, , (
Yi) can be obtained as position data of a rectangular portion including a figure.

Next, the CPU 8 connects the control terminals 11A and II of the selector 10.
A signal is sent to terminal B to select input terminal C of selector 10.

Next, CP[I8 sends the start signal 5OH (ISO code) to the register 13 via the terminal 12, R5 (ISO code) indicating 2-bit quantization, and then the position data L obtained as described above. Y+, △X1. △Y1 and the length Ll of the image data are sent to the selector 10.
The data is recorded on the disk memory 4 via the. The length of this pixel data can be found from: This length 1.1 is △Xl+
Since it can be calculated from ΔY1, it is not necessarily necessary to record it, but considering the operating speed during playback, it is preferable to record it in advance. Next, the CPU 8 sends control signals to the X and Y address generators 4 and 5, and sets the reading position of the frame memory 3 to the rectangle ABC obtained as described above.
The upper left point (XI, Yl) of 'D' is set, and the 8-bit quantized value at this point is read out and loaded into the 8-bit shift register 5, and then this shift register is shifted twice. Next, the 8-bit quantized value is loaded into the shift register 5 with the read address in the frame memory 3 as (L"l, L), and this is shifted again 2' times. If this operation is repeated 4 times, the shift The 8-bit shift register 16 connected in cascade with the register 5 stores the quantized values of the upper two bits of four pixels adjacent in the X-axis direction.
After selecting input terminal A of selector 10 via , 2 bits for 4 pixels stored in shift register 16!
The child value is read and recorded in the disk memory 14. By performing the above operation three times, the above rectangle ABC'D
The 2-bit quantized value of the pixel within ' will be recorded in the disk memory 14. In this case, there may be fractions, but since the W width is all black, they are naturally "0".
0". In this way, the first rectangle ABC
After recording the image information of the 'D' portion on the disk memo 1J14, all data within this rectangle in the frame memory 3 is erased as 0. Next, the above-mentioned operation is performed again to detect the second rectangle, and after recording the 2-bit quantized value of the pixels within the rectangle in the disk memory 14 together with the position data of this second rectangle, etc., the frame memory The data within this second rectangle is erased. These operations are repeated in sequence, and when the Y address reaches 485, scanning of the entire screen is completed, and the CP[J8 sends an End mark to the disk memory 14 and recording of one screen is completed. It turns out. As a result, the recording format in the disk memory 14 becomes as shown in FIG. This figure 4 shows a case where each image is a black and white image and a color image on the entire screen, but the same applies to the case where a color image and a black and white image are mixed in one screen. The data can be compressed and recorded on the disk memory 14.

In the above example, the image within the rectangle is a black and white image, and each pixel is converted into a 2-bit quantized value and stored in the disk memory 14.
As described above, when the image within the rectangle is a color image, input terminal B of the selector 10 is selected, the 8-bit quantized value read from the frame memory 3 is recorded in the register 17, and the selector 10 to the disk memory 14.

In this way, the conventional method of recording the information of the entire screen is done by extracting the part of the screen of the still image to be recorded that includes the image, and recording the pixel data of this part together with the position data representing this part. This means that the storage capacity of the disk memory 14 is only a few percent compared to the conventional method, and if the same capacity of disk memory is used, the number of screens that can be recorded is 2.
It becomes 5 to 150 times.

As mentioned above, sufficient resolution can be obtained by quantizing a super black and white image with 2 vias, but in the case of a black and white analog image, it is preferable to quantize with 4 bits. In this case, for example, 8
After storing the bit quantization value, the operation of shifting it four times is
After performing this for each pixel, the operation of recording the quantized values of the upper four bits of two pixels stored in the shift register 16 in the disk memory 14 may be performed in sequence. In this case, a signal GS (ISO code) indicating 4-bit quantization is added following the start signal SOH.

Next, the case of reproducing images from the disk memory 14 will be explained. As mentioned above, the image data recorded in the disk memory 14 includes a mixture of conventional 8-bit quantized still images, 2-bit quantized super images, and 4-bit 3-bit quantized images. Therefore, it is necessary to distinguish between these. The signal read from the disk memory 14 is stored in the register 21, and the signal following the synchronization signal SOH, US in the case of a color still image, R3 in the case of 2-bit quantization, and GS in the case of 4-bit quantization, is connected to the terminal 22. It is supplied to the CPU 8 via.

If the signal US indicating a color still image is detected, the 8-bit quantized value is stored as is in the frame memory 24 via the register 21 and the 8-bit shift register 23. In this case, terminal 25 has a C
Since “1” is given from P tl B, the animation
) 26-1 to 26-6 are all open and the frame memory 24
The 8-bit quantized value read from is directly supplied to the D/A converter 27, where it is converted into an analog signal.
It is output from the output terminal 28. The reproduction operation of this color still image is similar to the conventional method.

Next, the case where the signal R3 indicating 2-bit quantization is supplied from the terminal 22 to the CPU 8 will be explained (see 4-). Since a byte contains information on four pixels in series, one byte of data is first taken into the shift register 23 via the register 21.

Position data indicating a certain part of the image (XI
, Y, , ΔX, , ΔY+, L+) are supplied to the TEcP18 through the C terminal 22, so the CPt1 can know the position of this part in the screen. Based on this position data, the CPU 8 supplies address control signals via terminals 29 and 30 to X and Y address generators 31 and 32, respectively, each comprising a presettable counter, thereby generating frame 'J! h,)' Specifies the recording position within the memory 24. In this way, the 1 stored in the shift register 23 as described above is stored in the storage location of the frame memory 24 specified by the X and Y address generators 31 and 32.
After storing the 2-bit information for a pixel, it is shifted twice and the 2-bit information for the next pixel is stored in frames 1-, . 4
[-Stored in the 1st position following ri. Repeat this cycle once to 1-15 to store data for 4 pixels in the frame and memo U. In this case, a random value is stored in the lower 6 bits of each pixel position in the frame memory, but as mentioned above, since 26-1 to 26-6 are closed, their output is always There is no problem since it will be "0". Next, 1 byte of data is read again from the disk memory 14 and stored in the shift register 23, and the information of the upper 2 bits of the 4 pixels is written in a predetermined position in the frame memory 24 by the same operation as described above. ．． ! After reading bytes of data from the disk memory 14, the image data of the first rectangular portion is stored in the frame memory 24.
When the storage of the R3 code is completed and the R3 code is detected again, the second rectangular part is read out, and the same operation is repeated as [E
By repeating this process until the nd mark is detected, data for one screen is stored in the frame memory 24. Next, the image signal is reproduced at the output terminal 28 by supplying the signal read from the frame memory 24 to the D/A converter 27.

In the detailed explanation, it is assumed that one screen has only color images or black and white images, but color images and black and white images may be mixed. In this case, by detecting the codes LIS, R3, and GS recorded together with the position data of each part, it is possible to know what kind of data the part has. All you have to do is perform a playback operation.

Note that the gate signal for extracting the image read out from the )()-memory 24 can also be obtained by slicing the signal output from the D/A converter 27 at a predetermined level.
Particularly when there are many level steps, such as in 4-bit quantization, it is difficult to create a gate signal for extraction in this way.
The upper 2 bits (2
(in the case of quantization) or the upper 4 bits (in the case of 4-bit quantization) are supplied to the AND gate 33, and its output is used as a gate signal for extraction.

The present invention is not limited to the embodiments described above, but can be modified in many ways. In the above example, the coordinates (x+, y+) of the part containing the image were played back as they were, and the position of the reproduced image, 116 I-, was specified, but when re-'L, the - coordinates and coordinates were changed to f. You can move the relevant part to any position on the playback screen t by using -J", ". Many display screens can be easily created.

In this case, since only the data in each section is recorded, there is no need to increase the memory capacity by any means.

(Effects of the Invention) As described above, according to the image recording method of the present invention, when recording a mixture of color still images, black and white images such as super text and cut pictures on a disk memory, certain parts of the images The image information is recorded only for certain parts of the image, and it is determined whether the image is a color image or a monochrome image. By recording digitized information, the required memory capacity can be significantly reduced, and the system configuration can be made simple, compact, and inexpensive. Furthermore, during reproduction, by selectively reading out these color images and monochrome images, it is possible to easily construct an image in which color images and monochrome images coexist within one screen.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the overall configuration of an image recording and reproducing apparatus that implements the image recording method of the present invention, and FIG. Similarly, FIG. 3 is a diagram for explaining the method of determining the outline of the image at that time, and FIG. 4 is a diagram showing the recording format. 1... Input terminal 2... A/D converter 3...
・Frame memory 4.5...X, Y address converter 8...CPU 9...OR gate 1
0...Selector 13...Register 14...
・Disk memory 15.16...Shift register 2
1...Register 23...Shift register 2
4...Frame memory

Claims (1)

[Claims] 1. Divide the screen of the input image to be recorded into a portion with an image and a portion without an image, and for the portion with an image, a control signal representing the position and size of that portion on the screen;
The image information in this part is recorded, and further it is determined whether the input image is a color image or a monochrome image, and in the case of a monochrome image, the image information is quantized with a smaller number of bits than in the case of a color image. An image recording method characterized by recording.