JPS6314556B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a device for processing color image data at high speed.

Digital handling of color images has been carried out in various fields in recent years with the development of image processing technology using electronic computers and the like. First, the basic handling of color images will be explained with reference to FIG. 1 is an existing color television camera, and the captured image is separated into the three primary colors of red (R), green (G), and blue (B).
For each primary color image, for example, horizontal (X direction), vertical (Y direction)
(direction) into 640 and 480 pixels, respectively. The brightness and darkness of each pixel and each primary color is determined by the A/D converter 2 from 6 bits (64 gradations) to 8 bits from the analog signal.
It is converted into a bit (256 gradations) digital signal and one screen worth of data is stored in the image memory 3. The image memory 3 is a memory configured so that the value of each pixel can be read and written at the television scanning rate, and also serves as a refresh memory for the image monitor, and the read data is conversely D/A converted. By vessel 4,
Each primary color is converted into an analog signal and displayed on the color television monitor 5. Thereby, the image input by the television camera 1 can be confirmed directly with the naked eye. In addition, when processing input image data, a part of the data is read from the image memory 3 into a processing device 6 such as a computer, and arithmetic processing is performed. The processing results are returned to the image memory 3 and displayed on the color television monitor 5 to confirm the results with the naked eye.

However, with the basic structure above,
The drawback was that it could not support interactive image processing, which has become popular in recent years. For example, an object with a specific color of interest can be detected from an image on the color monitor 5.
When performing an operation to detect all object images in an image that have the same or similar color specified by the user above, each pixel is included in the color gamut [Equation (1)] specified by the user. The number of calculation instructions required to calculate R _l RR _u , G _l GG _u , B _l BB _u ...(1) is approximately 30 instructions, and if each instruction is executed in 1 μS, the number of calculation instructions is 640. ×
Calculation of 480 pixels takes approximately 9 minutes, excluding data transfer time.
It takes seconds. This response time is too long if the user wants to continuously change the color gamut using a tablet or the like and observe the distribution of the color in the image area with the naked eye. Furthermore, this condition is a fatal limiting condition when attempting to handle moving images. Therefore, a device that performs high-speed threshold processing as shown in equation (1) has a second
It may be installed as shown in the diagram. That is, a threshold circuit 7 is provided between the image memory 3 and the D/A converter 4.
If the pixel values (R, G, B) input from the image memory 3 satisfy equation (1), the threshold circuit 7 outputs a special emphasis color (for example, red) to the D/A converter 4. ) is output, and if not satisfied, the input (R, G, B) is output as is. Since the threshold data is controlled from the processing device,
It is independent of the image data flow and can be switched at high speed. The color image displayed on the television monitor 5 is constantly refreshed by the image memory 3, and pixels that satisfy equation (1) are highlighted in red by the threshold circuit 7, and this threshold value is changed in real time using a tablet or the like. Therefore, the user can obtain a desired color gamut while observing the result of color gamut change on the television monitor 5.

The conventional device as described above has the following drawbacks. First, when trying to recognize objects by color gamut, the colors of two objects are in R, G, and B space, as shown in 8 and 9 in Figure 3 shown in the RGB three-dimensional vector space diagram. If the color distribution of the object is 9, set the color gamut 10 as in equation (1) to
Even if you try to separate it from this axis, you cannot separate it from this axis. (8', 9', 10' are projections of 8, 9, 10 onto the RG plane) However, if the principal axes are transformed as in equation (2), instead of R, G, and B, α and β , γ may be recognized by applying threshold processing as shown in equation (1).

Second, when an image of a three-dimensional object is input using a television camera, even if the color of the surface is constant over the entire surface, the RGB three-dimensional vector space will look like the one shown in Figure 4 due to uneven lighting. Sometimes. That is,
As shown in the color distributions 11 and 12 of two objects, although the proportions of R, G, and B are constant, the colors are largely distributed in directions where the brightness differs.
As can be seen from the projections 11' and 12' of this distribution onto the RG plane, such a distribution is
Even if a threshold value is set as shown in equation (1) for
g) and brightness l, and can be separated by setting a threshold value as shown in equation (1) for (r, g, l).

r=R/R+G+B, g=G/R+G
+B, l=C ₁ R+C ₂ G+C ₃ B...(3) Executing the conversions shown in equations (2) and (3) above and using them for identification is as shown in Figure 2. It is possible to do this with conventional equipment. That is, the R, G, and B values of each point in the image memory 3 are converted in advance by the processing device 6, and the R, G, and B memories are filled with α, β, γ, or r, g,l
are stored, and the threshold value circuit 7 applies threshold value processing to each of them. However, in this case, it is necessary to superimpose the original color image on the TV monitor and obtain the desired threshold value in real time by superimposing the original color image on the TV monitor in the same way as for conventional R, G, B images. , since the data of B is not stored,
The original color image cannot be displayed. Also, if this was to be done, it would be necessary to prepare an image memory for one more frame.

The purpose of the present invention is to eliminate the above-mentioned drawbacks, and to perform threshold processing at high speed on data that has been transformed as shown in equations (2) and (3) without increasing the image memory capacity.
It is an object of the present invention to provide a device that can obtain the results in real time while preserving the original color image data.

Another object of the present invention is to provide color data that requires various conversions depending on the application, and which can be used for various applications with a single configuration, without configuring a separate conversion device. The purpose of the present invention is to provide a conversion device.

The present invention inputs digitized data signals of k (positive integer) color components, and first converts them into k
m first table converters, first adder circuits and m
(positive integer)+1 second table converters and n (positive integer including zero) auxiliary signal generation circuits generate auxiliary input signals and auxiliary output signals, and separately, k An output is obtained from the input by k×m third table converters, this output and the auxiliary input signal are added by m second adder circuits, and the output is added to m fourth table converters. By performing table conversion using the table converter group, m outputs and n auxiliary outputs can be obtained by setting appropriate values in the table converter group, and color data can be converted at high speed.

In the present invention, the calculations are only addition or subtraction by the addition circuit, and all other operations are based on table conversion, so it is very fast, and the conversion formula can be changed very flexibly. . In particular, by using the results of calculations on all inputs as auxiliary inputs for calculations on each output, the present invention has the advantage that most transformations appearing in color calculations are performed in real time at the high speed of television scanning.

The operation of the present invention will be explained below with reference to figures of embodiments. FIG. 5 shows the number of input signals of the present invention: 3 (k=
3), Number of output signals: 3 (m=3), Number of auxiliary output signals: 1
(n=1). The input signal is I ₁ ,
I ₂ , I ₃ , the output signals are O ₁ , O ₂ , O ₃ and the auxiliary output signal is E ₀ , respectively. Each input signal
I ₁ , I ₂ , I ₃ are each the first table converter 131
~ 133 converts the data into a table, enters the first addition circuit 14, and outputs the addition result signal F.
becomes auxiliary input signals E ₁ , E ₂ , E ₃ and auxiliary output signal E ₀ by the four second table converters 150 to 153. That is, 131-133,14,1
Each section 50 to 153 constitutes an auxiliary signal generation circuit. On the other hand, each input signal I ₁ , I ₂ , I ₃ is sent to 3×3=9 third table converter groups 1611 to 1613, 1
621-1623, 1631-1633 are converted separately, and the results and auxiliary input signals are
Three second adder circuits 1 are connected to each output in a matrix manner, with each input and auxiliary input all related.
7 input. The three outputs of the second adder circuit group 17 are connected to output signals O ₁ , O ₂ , O ₃ via three fourth table converter groups 181 to 183.

Incidentally, the table converters 131 to 131 appearing in the above explanation
133, 150-153, 1611-1613,
Each of 1621 to 1623 and 1631 to 1633 is a memory for table conversion, and the input signal (e.g. 8 bits) indicates the address (0 to 255) of the memory, and the output signal (e.g.
10 bits) is the content (0 to 1023) stored at that address that has been read out.
Since this conversion process is completed within the access time of the memory element, the conversion can be performed at a sufficiently high speed. Further, depending on what kind of table is stored in this memory, conversion can be performed using an arbitrary function of the form y=f(x) with one variable and one unknown.

It will be shown below that color data can be easily converted by giving a fixed function to the contents of the table converter. R, G, as shown in Figure 3.
When changing the color gamut by rotating the axis of the primary colors of B, a matrix operation is required as shown in equation (2), but in this case, inputs I ₁ , I ₂ , I ₃ are , the R, G, and B values are added. Table converter 1
The contents of address I of 611 are set to the value of _a11 ·I. Similarly, for the table converters 1612 to 1633, if the values of a ₁₂ · I to _{a 33} · I are set at address I, each adder circuit 17 calculates α,
β and γ can be obtained. In this case, since the table converters 181 to 183 are no longer needed, I is assigned to the I address as their contents. Output O ₁ ~
α, β, and γ appear in O ₃ .

When converting the tristimulus values of R, G, and B into chromaticity and brightness as shown in FIG. 4, calculation of equation (3) is required. Inputs I ₁ , I ₂ , I ₃ also have R, G,
The value of B is added. And table converter 1
I addresses 31, 132, and 133 have I addresses, respectively.
, and I is also placed at address I of the table converter 150. As a result, R+G+B of equation (3) is obtained for the signal F and the auxiliary output signal _E0 .
In the table converters 1611 and 1622, logI is placed at the I address. The other table converter 15
-logI is placed at address I of 1,152. At this time, in order not to disturb the calculation, the table converter 161
Zeros are placed in all addresses 2, 1613, 1621, and 1623. Furthermore, table converter 18
By placing a value of 10 ^I at address I of 1,182, r and g in equation (3) are obtained as outputs O ₁ and O ₂ .
(However, the scaling factor has been omitted for convenience of explanation.) Finally, l in equation (3) is the table converter 16
At addresses I of 31, 1632, and 1633, respectively.
It can be obtained by placing C ₁・I, C ₂・I, and C ₃・I. It is assumed that all addresses in the table converter 153 are set to zero, and the I address of 183 is set to I.

As explained above, according to the configuration of the present invention,
By rewriting the contents of each table converter, it is possible to convert data with a large degree of freedom. Even if the inputs are R, G, and B, there is also conversion to X, Y, and Z systems, generation of output including masking signals for printers, hue,
Conversions to purity, etc. can be performed at a speed that is sufficient to keep up with television data rates.

Next, how to use the present invention will be explained with reference to FIG.
In FIG. 6, in addition to the conventional configuration shown in FIG. 2, a color data conversion device 20 of the present invention is inserted between the image memory 3 and the threshold circuit 7. According to the color image analysis device having the configuration shown in FIG. 6, according to the present invention 20,
Desired color data converted from R, G, and B values using the functions set in each table converter is applied from the processing device 6 to the threshold circuit 7. However, in addition to the threshold value circuit 7 shown in FIG. 2, the threshold value signal 21 is generated when the threshold value set by the processing device 6 is satisfied, and the electronic switch 19 is controlled. The switch 19 receives R from the image memory 3,
D/A by switching between G, B signals and emphasized color signal
This is what is sent to the converter 4. With the above configuration, it is possible to easily analyze images that cannot be analyzed in real time using the conventional methods shown in FIGS. 3 and 4.

Although not shown in FIG. 6, the output signal of the color data conversion device 20 of the present invention is stored in the image memory
The present invention can also be used to reduce the burden on the processing device 6 by returning it to an image processing device. In this case, it goes without saying that E ₀ is required if the chromaticity and brightness are to be inversely converted back to (R, G, B) according to the present invention.

Furthermore, an embodiment will be shown in which a conversion unit that is a standardized version of the present invention is assembled. FIG. 7 shows an embodiment of the conversion unit 30. 4 inputs a,
b, c, and d are 8, 8, 8, and 10 bits, respectively, and the table converters 221 and 224 correspond to 256, 256, 256, and 102 bits, respectively.
It has 4 addresses. Table converter 221-2
Each of the 24 addresses has a capacity of 10 bits. The conversion results are added by an adding circuit 23. The addition result is obtained in the form of 10 bits at the output ξ, which is converted by the table converter 24 into an 8-bit value and obtained at the output η. FIG. 8 shows an example in which the embodiment of FIG. 5 is realized using four conversion units 300, 301, 302, and 303. In the examples shown in Figs. 7 and 8, the same conversion unit was used, so the 5th
Compared to the example in the figure, the number of table converters is one extra, and correspondingly, the number of extra inputs is
Ex is increasing. However, the same effect can be achieved by placing zeros in all addresses of the table converter. Moreover, furthermore,
When setting the number of input signals k, the number of output signals m, and the number of auxiliary signals n to arbitrary numbers, for example, k=3, m=4, n=
1, the conversion unit in Figure 7 is m+n=4+
1=5, of which 4 are for output signals and 1 is for auxiliary signals, and can be easily obtained by connecting them in the same manner as shown in FIG. Also, in the case of k=3, m=2, and n=1, the same result can be obtained using m+n=2+1=3 conversion units. Furthermore, k=3, m=3, n=2
, k = 4, m = 3, n = 2, this can be easily realized based on the previous explanation by arranging the input side table converters shown in Fig. 7 according to the numbers of k and m. It is something. As a result, by integrating and manufacturing each unit, the same manufacturing process can be used for various uses, which has the effect of lowering costs.

In the above description, the three primary color signals of R, G, and B are used as input signals of the present invention, but the effects of the present invention are not limited thereto; for example, r, g, l, E ₀ Converting the chromaticity, brightness and auxiliary output signals back to R, G, B, CIE
(International Commission on Illumination) Established in 1931, it can be used to convert many color data, such as rapidly converting the X, Y, Z tristimulus values to other color matching functions.

[Brief explanation of the drawing]

Figures 1 and 2 are block diagrams showing the configuration of a conventional color image analysis device, and Figures 3 and 4 are block diagrams showing the structure of a conventional color image analysis device.
An RGB three-dimensional vector space diagram, FIG. 5 is a block diagram showing an embodiment of the present invention, FIG. 6 is a block diagram showing the configuration of a color image analysis device including the present invention in its configuration, and FIGS. FIG. 8 is a block diagram showing an embodiment of the present invention constituted by several conversion units. In the figure, 1 is a color television camera, 2
is an A/D converter, 3 is an image memory, 4 is a D/A converter, 5 is a color TV monitor, 6 is a processing device,
7 is a threshold circuit, 131-133, 150-15
3, 1611 to 1633 and 181 to 183 are table converters, 14 and 17 are addition circuits, I ₁ to _{I 3}
is an input signal, O ₁ to O ₃ are output signals, E ₀ is an auxiliary output signal, 20 is a color data converter, 30 and 30
0 to 303 are conversion units.

Claims (1)

[Claims] 1 Color data conversion that inputs k (positive integer) digitized color data signals and obtains m (positive integer) outputs and n (positive integer including zero) auxiliary outputs. The device includes k first table converters that convert input data corresponding to each of the k inputs, a first addition circuit that adds all of the conversion results, and a table conversion of the addition results again. , m auxiliary input signals and 1 corresponding to the output
n auxiliary signal generation circuits having m+1 second table converters that generate auxiliary output signals;
k×m third table converters that separately convert each of the k input signals corresponding to the m output signals; and k third table converters corresponding to the m outputs. m second adder circuits that add the outputs of and the n auxiliary input signals, and m that converts the output of the second adder circuits into a table again for each output.
a fourth table converter, and performs data conversion at high speed by assigning a preset value to each table converter group.