JP5757108B2 - Image processing apparatus and program - Google Patents

Description

  The present invention relates to an image processing apparatus and a program.

  Patent Document 1 includes a table memory that stores a numerical value obtained by multiplying a square value of a difference between two colors of three colors of a color image signal by a coefficient, and an arithmetic unit that calculates an absolute value of the difference between the two colors. There is disclosed a masking operation device that calculates a quadratic term of a masking operation expression by referring to a table memory using an output of an operation unit as a read address.

JP-A-62-293344

  The object of the present invention is less than the case where shift operations are performed on all terms after multiplication by transform coefficients and the case where shift operations are performed collectively after adding all terms multiplied by transform coefficients. An object of the present invention is to provide an image processing apparatus and a program capable of realizing color conversion with high accuracy by arithmetic processing.

[Image processing device]
According to the first aspect of the present invention, a plurality of values of a multi-order term obtained by multiplying image data of the first color system expressed by a plurality of input color components by a combination of a plurality of color components. A calculating means for calculating;
A plurality of first shift means for shifting the values of a plurality of multi-order terms calculated by the calculation means by a preset number of bits;
A plurality of preset conversion coefficients are set for each color component of the second color system for the input values of the plurality of color components and the values after being shifted by the first shift means. A plurality of multiplication means for multiplying;
A plurality of addition means for adding the values after the conversion coefficients are multiplied by the plurality of multiplication means for each color component of the second color system;
A plurality of second shift means for shifting the values added by the plurality of addition means by a predetermined number of bits and outputting the values as the values of the respective color components of the second color system image data;
Is an image processing apparatus.

  The present invention according to claim 2 is the image processing apparatus according to claim 1, wherein the first shift means shifts the values of the plurality of multi-order terms by a number of bits corresponding to the order of the multi-order terms.

  According to a third aspect of the present invention, there is provided a multiplication value of two different color components selected from the plurality of color components and a square value of each color component by the calculation means for the image data of the first color system. The image processing apparatus according to claim 1, wherein each is calculated as a value of the multi-order term.

  According to a fourth aspect of the present invention, the image data of the first color system is image data of an RGB color system, and the image data of the second color system is an L * a * b * color system. The image processing apparatus according to any one of claims 1 to 3, wherein the image data is image data.

[program]
According to the fifth aspect of the present invention, a plurality of values of a multi-order term obtained by multiplying the first color system image data expressed by a plurality of input color components by a combination of a plurality of color components are provided. A calculating step;
A shift step for shifting the calculated values of the multiple-order terms by a preset number of bits,
For values after being shifted by the value and the previous carboxymethyl shift steps of the input of the plurality of color components, a plurality of conversion coefficients set in advance, respectively multiplying each color component of the second color system And steps to
Adding the value after the multiplication by the conversion coefficient for each color component of the second color system;
A program for causing a computer to execute a step of shifting a value added for each color component by a predetermined number of bits and outputting the value as a value of each color component of second color system image data .

  According to the first aspect of the present invention, the shift operation is performed on all the terms after multiplication by the transform coefficient, or the shift operation is performed collectively after all the terms multiplied by the transform coefficient are added. In comparison, it is possible to provide an image processing apparatus capable of realizing highly accurate color conversion with less arithmetic processing.

  According to the second aspect of the present invention, in addition to the effect obtained by the first aspect of the present invention, it is possible to obtain the effect that the balance between the calculation processing amount and the color conversion accuracy can be appropriately maintained.

  According to the third aspect of the present invention, when the secondary term is calculated from the image data of the first color system and color conversion is performed, the shift operation is performed on all the terms after multiplying the conversion coefficient. To provide an image processing apparatus capable of realizing color conversion with high accuracy with a small amount of arithmetic processing, compared to a case where shift operations are performed collectively after adding all terms multiplied by conversion coefficients. Can do.

  According to the fourth aspect of the present invention, when color conversion is performed on image data of the RGB color system to image data of the L * a * b * color system, all the terms are applied after multiplying the conversion coefficient. Compared to the case where shift calculation is performed, or the case where shift operation is performed collectively after adding all the terms multiplied by the conversion coefficient, image processing that can realize highly accurate color conversion with less calculation processing An apparatus can be provided.

  According to the fifth aspect of the present invention, the shift operation is performed on all the terms after multiplication by the transform coefficient, or the shift operation is performed collectively after all the terms multiplied by the transform coefficient are added. In comparison, it is possible to provide a program capable of realizing highly accurate color conversion with a small amount of arithmetic processing.

It is a figure for demonstrating the 3 * 9 matrix calculation performed in order to convert the image data of RGB color system represented by the RGB color component into image data of L * a * b * color system. It is a figure which shows the example of the value of the conversion coefficient in the case of performing matrix operation as shown in FIG. FIG. 3 is a diagram illustrating a configuration of an image processing apparatus for performing arithmetic processing for converting RGB color system image data into L * a * b * color system image data by digital processing. It is a figure which shows an example of a specific structure of the calculator 82 of a normal system. FIG. 5 is a diagram illustrating an example of a conversion coefficient and a shift amount used in the normal-type computing unit 82 illustrated in FIG. 4. FIG. 5 is a diagram for specifically explaining multiplication processing and shift calculation processing in the normal-type computing unit 82 shown in FIG. 4. It is a figure which shows an example of a specific structure of the arithmetic unit 82 of a common shift system. It is a figure which shows the example of the conversion coefficient and common shift amount which are used in the arithmetic unit 82 of the common shift system shown in FIG. It is a figure which shows the hardware constitutions of the image processing apparatus of one Embodiment of this invention. It is a block diagram which shows the function structure of the image processing apparatus of one Embodiment of this invention. It is a figure which shows the example of the conversion coefficient used in the image processing apparatus of one Embodiment of this invention, the preshift amount, and the common shift amount. It is a figure which shows the calculation result in a real number calculation value, a normal system, a common shift system, and the calculation system of this embodiment. It is the figure which described the arithmetic processing at the time of using the DSP provided with the function which can perform a shift arithmetic processing and a multiplication process simultaneously in 1 clock by a circuit expression.

[background]
First, in order to help understanding of the present invention, its background and outline will be described.

  As a method for converting image data of a certain color space into image data of another color space, a method of performing an N × M matrix operation is known. For example, in order to convert RGB color system image data represented by RGB color components into L * a * b * color system image data, a 3 × 9 matrix operation as shown in FIG. 1 is performed. The method is known.

  In the matrix operation shown in FIG. 1, nine values B, G, R, B × B, G × G, R × R, B × G, G × R, and R × B are input as input signals. As an output, three values L *, a *, and b * are obtained as output signals.

  By performing this matrix operation, for example, an operation such as L * = C11 × B + C12 × G +... + C19 × R × B is performed, and the values of L *, a *, and b * are calculated. . FIG. 2 shows an example of the values of the conversion coefficients C11, C12,... C39 when performing such a matrix operation.

  In order to perform the arithmetic processing as described above by digital processing, generally, first convert the real value of the conversion coefficient into an integer value and a shift amount. A method of performing division by multiplying an integer value and shifting the obtained value by a preset shift amount is performed.

  FIG. 3 shows the configuration of an image processing apparatus for performing such calculations. This image processing apparatus includes a secondary term generation unit 81 and calculators 82, 83, and 84 provided for each color component of L *, a *, and b *.

  The secondary term generation unit 81 inputs RGB color system image data expressed by the three color components B, G, and R, and selects all the two color components that can be selected from the three color components. Multiplication values B × G, G × R, R × G and square values (square values) B × B, G × G, R × R of the respective color components are calculated and output together with the input three color components. .

  The calculators 82, 83, and 84 calculate L *, a *, and b * values based on the nine values output from the color component generation period 81, respectively.

  An example of a specific configuration of the calculator 82 is shown in FIG. The calculator 82 shown in FIG. 4 is the most general method (hereinafter referred to as a normal method), and includes multipliers 91 to 99, shift calculators 101 to 109, and an adder 110. .

  Multipliers 91 to 99 are respectively provided for the nine values B, G, R, B × B, G × G, R × R, B × G, G × R, and R × B as input values. Multiplication coefficients D11, D12,..., D19 as shown in FIG.

  5, conversion coefficients D11,..., D39 and shift amounts sft11,..., Sft39 shown in FIG. 5 are obtained by converting the conversion coefficients shown in FIG.

  Then, the shift calculators 101 to 109 are values obtained by multiplying the conversion coefficients by the multipliers 91 to 99 by the number of bits specified by the shift amounts sft11, sft12,... Sft19 as shown in FIG. To shift.

  The adder 110 adds all the output values from the shift calculators 101 to 109 and outputs the result as an L * value.

  5, conversion coefficients D21, D22,..., D29 and shift amounts sft21, sft22,..., Sft29 are for calculating the value of a *. The conversion coefficients D31, D32,..., D39 and the shift amounts sft31, sft32,..., Sft39 are for calculating the value of b *.

  In such a color conversion process, the multiplication process and the shift operation process are performed as shown in FIG. FIG. 6 shows a case where the value of the color component B is multiplied by a conversion coefficient D11 = −571 and a shift operation is performed by sft11 = 12 bits.

  Note that the value of the color component B that is an input value is expressed by 8 bits, the conversion coefficient D11 is expressed by 11 bits with a sign, and the shift amount sft11 is expressed by 5 bits.

  The input value 100 is multiplied by the conversion coefficient D11 = −571 by the multiplier 91 to become a value (−57100) expressed in 19 bits. This value is shifted by 12 bits by the shift computing unit 101 to become −13.9375, and is output to the adder 110.

  If the real number calculation is performed using the conversion coefficient C11 = −0.139610126615 shown in FIG. 2, since 100 × C11 = −13.96610126615, almost the same result can be obtained by the method shown in FIG. You can see that

  Another example of the specific configuration of the computing unit 82 is shown in FIG. The arithmetic unit 82 shown in FIG. 7 has a configuration called a common shift method in which the shift operation is performed only once on the output value of the adder 110 for the purpose of reducing the number of times the shift operation is performed. , Multipliers 91 to 99, an adder 110, and a shift calculator 120. In FIG. 7, the same components as those in FIG. 4 are denoted by the same reference numerals, and the description thereof is omitted.

  Multipliers 91 to 99 are respectively provided for the nine values B, G, R, B × B, G × G, R × R, B × G, G × R, and R × B as input values. Multiplication coefficients E11, E12,..., E19 as shown in FIG.

  The conversion coefficients E11,..., E39 and the common shift amounts sftL, sfta, sftb shown in FIG. 8 are obtained by converting the conversion coefficients shown in FIG. 2 into integer conversion coefficients and common shift amounts.

  The adder 110 adds all the output values from the multipliers 91 to 99 and outputs the result. The shift computing unit 120 shifts the output value of the adder 110 by the number of bits specified by the common shift amount sftL (14 bits) and outputs it as a value of L *.

  According to this common shift method arithmetic processing, it can be seen that the shift arithmetic processing is reduced as compared with the normal method of FIG.

[Embodiment]
Next, embodiments of the present invention will be described in detail with reference to the drawings.
FIG. 9 is a diagram showing a hardware configuration of the image processing apparatus according to the embodiment of the present invention.

  As shown in FIG. 9, the image processing apparatus according to the present embodiment includes a DSP (Digital Signal Processor) 11, a memory 12, and a ROM 13.

  The DSP 11 executes a predetermined process based on a control program stored in the memory 12 or the ROM 13 to perform color conversion processing of RGB color system image data to L * a * b * color system image data. Realize. In the present embodiment, the DSP 11 has been described as reading and executing a control program stored in the memory 12 or the ROM 13, but the program is stored in a storage medium such as a CD-ROM and provided to the DSP 11. It is also possible.

  FIG. 10 is a block diagram illustrating a functional configuration of the image processing apparatus according to the present exemplary embodiment that is realized by executing the control program.

  As shown in FIG. 10, the image processing apparatus according to the present embodiment includes a secondary term generation unit 20, a pre-shift unit 30, multiplication units 40L, 40a, and 40b, adders 50L, 50a, and 50b, and a shift calculator. 60L, 60a, and 60b.

  The preshift unit 30 includes shift calculators 31 to 36. The multiplication unit 40L includes multipliers 41 to 49. Note that the configuration of the multipliers 40a and 40b is the same as that of the multiplier 40L, and a description thereof will be omitted.

  The quadratic term generation unit 20 multiplies the values (BG, GR, GB) of two different color components selected from the three color components R, G, B with respect to the RGB color system image data and the square of each color component. The values (BB, GG, RR) are respectively calculated as secondary term values. Note that the secondary term generation unit 20 in the present embodiment also outputs the value of each color component of the input BGR as it is.

  The shift calculators 31 to 36 in the preshift unit 30 respectively set the values of the six secondary terms calculated by the secondary term generation unit 20 to the number of bits set by the preshift amounts Sbb, Sgg,. shift.

  Multipliers 41 to 49 in the multiplier 40L are preset with respect to the values (B, G, R) of the three color components of the RGB color system image data and the values after the shift operation is performed in the preshift unit 30. The nine conversion coefficients C1 to C9 are multiplied for each color component of the L * a * b * color system.

  The adder 50L adds and outputs the values after the multiplication coefficients C1 to C9 are multiplied by the nine multipliers 41 to 49.

  The shift computing unit 60L shifts the value added by the adder 50L by the number of bits set as the common shift amount SL, and outputs it as the value of the L * component of the L * a * b * color system image data. To do.

  In addition, by the same process as described above, the value of the a * component is calculated by the multiplier 40a and the adder 50a, and the value of the b * component is calculated by the multiplier 40b and the adder 50b.

  In this embodiment, the number of shift operations necessary to obtain output values of three colors is a total of nine times performed in the shift calculators 31 to 36, 60L, 60a, and 60b.

  On the other hand, the normal method shown in FIG. 4 requires nine shift arithmetic processes 101 to 109 in order to obtain one color output, and in order to obtain three color outputs. 27 (= 9 times × 3 colors) shift operations are required.

  In the present embodiment, the second-order terms (BB, GG, RR, BG, GR, RB) are subjected to a shift operation in the pre-shift unit 30 before being multiplied by the conversion coefficients C4 to C9. , B *, it is possible to perform a common shift operation, and the number of shift operations is reduced compared to the case where the shift operation is performed independently for each color.

  Next, FIG. 11 shows examples of conversion coefficients C1 to C9, preshift amounts Sbb to Srb, and common shift amounts SL, Sa, and Sb used in the image processing apparatus of the present embodiment. The conversion coefficient, pre-shift amount, and common shift amount shown in FIG. 11 are values set to perform the same calculation as the calculation using the conversion coefficient shown in FIG. Here, the conversion coefficient for calculating the L * component, the conversion coefficient for calculating the a * component, and the conversion coefficient for calculating the b * component are all expressed by C1 to C9. As shown in FIG. 5, these conversion coefficients are different coefficients set for each color of L *, a *, and b *.

  In the example shown in FIG. 11, the preshift amounts Sbb to Srb are commonly set to 4 bits, and the common shift amounts SL, Sa, and Sb are all set to 14 bits. It is also possible to set different values.

In order to replace the real value as shown in FIG. 2 with the integer part and the shift amount, first, the largest value in each real value is kept to the maximum within the range not exceeding the bit width of the integer part. To do. For example, when the conversion coefficient shown in FIG. 2 is replaced with an integer part and a shift amount, if the converted conversion coefficient is expressed with signed 16 bits, the conversion coefficient must be within a range of ± 32768 (= 2 ¹⁵ ). Don't be. Since the conversion coefficient having the largest absolute value among the conversion coefficients shown in FIG. 2 is C22 = 1.395506629967, a common shift amount is set such that the conversion coefficient C22 falls within the above-described range. Specifically, since 1.395056696967 × 2 ¹⁴ = 22856.661326 and 1.395056696967 × 2 ¹⁵ = 45713.22652, 14 is set as the common shift amount not exceeding the above range.

Then, the pre-shift amount of the pre-shift unit 30 is set so that the conversion coefficient for the second-order term becomes large within a range not exceeding the bit width of the integer part. For example, when the pre-shift amount of 4 bits, the conversion coefficient C4~C9 for the second-order term, set to a value of about 2 ⁴ times.

  Therefore, as compared with the conversion coefficient of the common shift method shown in FIG. 8, it can be seen that the conversion coefficient in this embodiment shown in FIG. . Therefore, as a result of the arithmetic processing in this embodiment, the effective bits of the conversion coefficients C4 to C9 multiplied by the multi-order term are larger than those in the common shift method.

  This is because the second-order term is a multiplication value of two values, so that the absolute value is larger than the RGB single value and the conversion coefficient is relatively small. This is because an error between the calculated value and the real number is reduced by increasing the quadratic term.

  For example, FIG. 12 shows the calculation results by each calculation method when the input values are B = 150, G = 200, and R = 250.

  Referring to FIG. 12, it is the normal method that obtains the closest value compared to the real number calculated value (theoretical value), but the calculation result in the present embodiment also has almost the same accuracy as this normal method. I understand that In addition, it can be seen that the color difference between the calculation result of the common shift method and the real number calculation value is larger than that of the normal method or the calculation method in the present embodiment.

  Also, when a DSP having a function capable of simultaneously executing a shift operation process and a multiplication process in one clock is used, a preshift operation process (preshift unit 30) for a second-order term and a multiplication process in the multiplier unit 40L are executed in one clock. It becomes possible to do. That is, by using such a DSP, the color conversion accuracy approaches that of the normal method with the same amount of calculation processing as the common shift method shown in FIG.

FIG. 13 shows the arithmetic processing in the case of using such a DSP in a circuit expression.
In the configuration shown in FIG. 13, arithmetic processing is performed by shift calculators 31 to 36, multipliers 41 to 49, adders 71 to 78, and shift calculator 60L.

  As shown in FIG. 13, if a DSP having a function capable of simultaneously executing a shift calculation process and a multiplication process in one clock is used, the processes of the shift calculator 31 and the multiplier 44, the shift calculator 32, the multiplier 45, and The processing of the adder 71 and the processing of the shift calculator 36, the multiplier 49, and the adder 75 are executed in one clock.

[Modification]
In the above-described embodiment, in order to simplify the description, a quadratic term (BB, GG, RR, BG, GR, RB) is calculated from the RGB three-color color system image data, and matrix calculation for color conversion is performed. However, the present invention is also applicable to a case where color conversion is performed by calculating a third-order term or higher order term.

  For example, when the input image data is RGB color system image data, 10 terms of BBB, GGG, RRR, BBR, BBG, GGB, GGR, RRB, RRG, BGR are calculated as the third-order terms. It is possible to realize color conversion processing by performing a matrix operation together with the second-order term and the value of each color component described.

  In this case, in the configuration shown in FIG. 10, the secondary term generation unit 20 uses the multi-term terms obtained by multiplying the color system image data represented by the plurality of color components by a combination of the plurality of color components. What is necessary is just to replace with the multi-order term production | generation part which calculates multiple values. Then, the pre-shift unit 30 may shift the calculated values of the multiple-order terms by a preset number of bits.

  Further, in such a configuration, the preshift unit 30 may shift the values of a plurality of multi-order terms by the number of bits corresponding to the order of the multi-order terms. That is, the shift amount for performing the shift operation may be set for each order of the multi-order term.

  Further, in the above-described embodiment, 3-input 3-output color conversion between the RGB color space and the L * a * b * color space (color conversion for converting three color components into three color components). However, the present invention is not limited to this, and a plurality of colors including other color spaces such as RGB, CMY, CMYK, L * a * b *, YCbCr, and XYZ. The present invention can be similarly applied even when color conversion is performed between any two color spaces in the space.

  However, for example, when performing 3-input 4-output color conversion, new conversion coefficients C41 to C49 are required for the conversion coefficients shown in FIG.

  When performing color conversion with four inputs and three outputs, the number of input terms increases by the value of the fourth color, the square of the fourth color, and the term of the combination of the fourth color and the other three colors. For example, when CMYK is input, C, M, Y, K, C × C, M × M, Y × Y, K × K, C × M, M × Y, Y × C, K × C , K × M, and K × Y are 14 terms. Further, as the number of input terms increases, conversion coefficients corresponding to the increased input terms are also required.

  Further, when performing 4-input 4-output color conversion such as color conversion from the CMYK color space to the C′M′Y′K ′ color space, it can be realized by combining the above two cases. is there.

11 DSP
12 memory 13 ROM
20 Second-order term generator 30 Pre-shift unit 31-36 Shift calculator 40L, 40a, 40b Multiplier 41-49 Multiplier 50L, 50a, 50b Adder 60L, 60a, 60b Shift calculator 71-78 Adder 81 Secondary term generator Units 82 to 84 arithmetic units 91 to 99 multipliers 101 to 109 shift arithmetic units 110 adders 120 shift arithmetic units

Claims (5)

Calculating means for calculating a plurality of values of multi-order terms obtained by multiplying a combination of a plurality of color components with respect to the first color system image data expressed by the plurality of input color components;
A plurality of first shift means for shifting the values of a plurality of multi-order terms calculated by the calculation means by a preset number of bits;
A plurality of preset conversion coefficients are set for each color component of the second color system for the input values of the plurality of color components and the values after being shifted by the first shift means. A plurality of multiplication means for multiplying;
A plurality of addition means for adding the values after the conversion coefficients are multiplied by the plurality of multiplication means for each color component of the second color system;
A plurality of second shift means for shifting the values added by the plurality of addition means by a predetermined number of bits and outputting the values as the values of the respective color components of the second color system image data;
An image processing apparatus.   The image processing apparatus according to claim 1, wherein the first shift unit shifts the values of the plurality of multi-order terms by a number of bits corresponding to the order of the multi-order terms.   The calculation means calculates a multiplication value of two different color components selected from the plurality of color components and a square value of each color component as the values of the multi-order terms for the first color system image data. The image processing apparatus according to claim 1.   The first color system image data is RGB color system image data, and the second color system image data is L * a * b * color system image data. 4. The image processing apparatus according to any one of items 3. Calculating a plurality of values of multi-order terms obtained by multiplying image data of the first color system expressed by a plurality of input color components by a combination of a plurality of color components;
A shift step for shifting the calculated values of the multiple-order terms by a preset number of bits,
For values after being shifted by the value and the previous carboxymethyl shift steps of the input of the plurality of color components, a plurality of conversion coefficients set in advance, respectively multiplying each color component of the second color system And steps to
Adding the value after the multiplication by the conversion coefficient for each color component of the second color system;
A program for causing a computer to execute a step of shifting a value added for each color component by a predetermined number of bits and outputting the value as a value of each color component of second color system image data.

Images