JPH0624005B2 - Gradation conversion circuit using lookup table - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a technique for reducing tone jumps, false contours, etc. of a gradation conversion circuit in image signal processing, particularly by a digital conversion method.

<Prior Art> In a scanner, a general image input device, or an image processing device, some gradation conversion circuit (gradation circuit,
The gamma correction circuit) is indispensable.

The gradation conversion circuit generally has a function f for input data x.
To obtain output data y = f (x), which is realized by both an analog circuit and a digital circuit. The input data x and the output data y are both data relating to density.

The analog method cannot obtain arbitrary function characteristics and has problems in response speed and stability. However, the digital method uses a so-called look-up table (LUT) to convert input data into memory addresses. , Output data of memory of arbitrary function can be obtained.

Also, the processing speed is determined by the memory access time,
It solves problems with analog circuits such as high speed and high stability. The memory can be RAM (random access memory) or ROM (read only memory).
However, if the RAM is used, it becomes easy to change the function characteristics.

<Problems to be Solved by the Invention> However, a digital gradation conversion circuit using a look-up table has a problem of quantization error which is essentially not a problem in an analog method. That is, since the gradation conversion circuit or the like creates a gradation curve having a non-linear characteristic with a look-up table, the quantization error of the output data becomes large.

For simplicity of explanation, the input / output data x, y
Fig. 77 gives a simple example in which is 4 bits. It is usually 8 bits, and a smoother curve is drawn as the number of pits increases.

In FIG. 7, it is assumed that the input / output conversion characteristic for the output target value y with respect to the input x is a curve A indicated by a dotted line. However, since it is digital in reality, it has a stepwise characteristic like a curve B shown by a thick line. For example, x =
When 6, the output data y = 9.3 for the target characteristic of curve A
However, since y is digital, the curve B is rounded to y = 9. Obviously, such a stepwise curve B is not always preferable for tone reproduction.

More specifically, when the input data x changes from 0 to 1, the output data y suddenly changes from 0 to 3. Even when x changes from 2 to 3, y changes from 4 to 6. That is, an inconvenient phenomenon called so-called tone jump or false contour in which gradation suddenly changes occurs.

On the contrary, at x = 9,10, y remains 12 and does not change. The same applies to x = 12 and 13. In such a place, a subtle change in gradation cannot be reproduced.

As a solution, it is conceivable to reduce the error by increasing the number of bits of the data x and y, but the increase of the number of bits leads to the complexity of the circuit and the cost increase. There is.

Therefore, in order to reduce the quantization error, conventionally, as shown in FIG. 8 (A) or (B), a method of mixing random numbers (noise) in the data to make the influence of the quantization error inconspicuous can be considered. ing.

In the figure (A), the random number data randomly output from the random number generation circuit 21 and the input data x are added by the adder 22, and the address of the lookup table LUT is designated by the added data, and the designated address is designated. The output data y is configured to be output in correspondence with the above.

Further, FIG. 6B shows random number data randomly output from the random number generation circuit 21 with respect to the data output from the lookup table LUT as a result of designating the address of the lookup table LUT by the input data x. Addition is performed by the adder 22, and the addition result is output data y.

However, in these methods, noise is directly mixed in the signal, and therefore it may not be preferable in reducing tone jumps and false contours.

An object of the present invention is to solve the above-mentioned problems without increasing the number of bits of a pixel signal and preventing the adverse effect of direct mixing of noise.

<Means for Solving Problems> In the present invention, the look-up table itself is devised so as to have randomness, rather than mixing noise into the signal. That is, instead of using a fixed lookup table with only one input / output conversion characteristic, prepare multiple types of lookup tables with different input / output conversion characteristics and randomly select the lookup table to be used. It is the one.

The gradation conversion circuit using the look-up table of the present invention based on such an idea outputs N conversion data (N is 2 or more) stored in advance when digital image input data and a selection signal are input. and a look-up table of integers) of the said selection signal for selecting one of the N lookup tables and a random number generating circuit for outputting the random input data x _{j (i} continuous Target value y _j of the converted data corresponding to the integer part I _j and the decimal part α _j (y _j = I
_j + α _j ; 0 ≦ α _j <1), (1-α _j ) · N or the closest integer number of conversion data of the lookup tables are set to y _j = I _j , and N out of N Of the remaining lookup table conversion data of y _j = I _j +1
It is characterized in that the input / output conversion characteristics of each of the N look-up tables are defined based on the above rule.

In this configuration, the random number generation circuit is in a broad sense, and includes a circuit for generating a so-called random number, a pseudo random number generation circuit, and a pattern generation circuit for storing a specific pattern and reading it cyclically. .

Note that this configuration includes at least the following two modes.

One is that N look-up tables are formed in respective memories, and the selection signal is configured to select any one of the N memories.

The other one is configured such that N look-up tables are formed in a single memory, and when the input data and the selection signal are input as the address signal, the conversion data of the corresponding address is output. It was done.

<Operation> The operation of the configuration of the present invention is as follows.

When (1-α _j ) · N is not an integer, y _j = I _j
Number of lookup tables to be converted to (1-α _j ).
N is rounded off or rounded down to (1-α _j ) · N, which is an integer number n _j (≈ (1-α _j ) · N), and y _j =
The number of lookup tables to be converted into I _j +1 is m _j
= N−n _j (≈α _j · N).

A certain lookup table is selected by the selection signal output from the random number generation circuit in synchronization with the input of the input data x _j . The look-up table is either a look-up table for converting input data x _j into output data y _j = I _j or a look-up table for converting output data y _j = I _j +1.

The probability that the selection signal from the random number generator selects the former lookup table (x _j → y _j = I _j ) is n _j / N.
≈ (1−α _j ) · N / N = 1−α _j , and the probability of selecting the latter lookup table (x _j → y _j = I _j +1) is m _j / N≈α _j · N / N = α _j .

That is, the output data y _j is substantially (1-α
y _j = I _j is output with a probability of _{j j} ) and y _j = I _j +1 is output with a probability of α _j .

Accordingly, the average value of the output data _{y j} (= expected _{value) j}
Is _j = (1-α _j ) I _j + α _j (I _j +1) I _j +
α _j , which matches the target value.

<Example> Hereinafter, an example of the present invention is described in detail based on a drawing.

First Embodiment FIG. 1 is a block diagram of a gradation conversion circuit using a lookup table according to the first embodiment, FIG. 2 is a diagram showing a relationship between input data and a target value, and FIG. 3 is a gradation conversion circuit. FIG. 4 is a schematic diagram showing input / output conversion characteristics of the lookup table in FIG. 4, and FIG. 4 is a schematic block diagram of a color scanner using the gradation conversion circuit.

In FIG. 4, 1 is a photoelectric conversion element that optically reads the density of the original image in pixel units and converts it into a current amount, 2 is an I / V conversion circuit that converts current into voltage, and 3 is a voltage amplifier. Three such circuits are provided for the three primary colors of R, G, B (red, green, blue) light. Immediately before each photoelectric conversion element 1, R, G, and B filters are arranged.

4 is an R signal, a G signal input from the three voltage amplifiers 3,
A color operation circuit (analog type) for converting the B signal into each signal of Y, M, C, K (yellow, magenta, cyan, black) which is a printing color, 5 is an A / D conversion circuit, and 6 is a magnification conversion circuit. , 7 are gradation conversion circuits according to the first embodiment of the present invention, and 8
Is a dot generator for creating halftone dots, and 9 is an exposure head.

As shown in FIG. 1, the gradation conversion circuit 7 includes N look-up table LUTs having different input / output conversion characteristics.
_{1 to} LUT _N and these N look-up tables L
A random number generation circuit 10 for individually outputting a selection signal S _i for randomly selecting one lookup table LUT _i from UT _{1 to} LUT _N as random number data to each lookup table LUT _i. ing. The random number generation circuit 10 updates the random number data to be output each time a clock corresponding to a pixel signal corresponding to one pixel is input. Each lookup table LUT _{1 to} LUT _N is R
It is formed in the OM or the RAM.

The input data x is a plurality of lookup tables LUT ₁
Although inputted to ～LUT _N, the plurality of look-up tables _LUT 1 ~LUT _N is any one is selected by the selection signals _{S 1} from the random number generation circuit 10. Then, from the selected look-up table LUT _i, the output data y in accordance with the input-output conversion characteristic that is previously stored in the look-up table LUT _i is output.

Next, N look-up tables LUT _{1 to} LUT _N
A method of creating the input / output conversion characteristic of will be described.

In FIG. 7, as described above, when x = 6, the output characteristic y of the curve A is y = 9.3, but since it is digital, it is rounded to y = 9. This is curve B.

Under such a condition, in the present invention, when the input / output conversion characteristics of the _N lookup tables LUT _{1 to} LUT N are created, the value of y when x = 6 is set to 9 when there is,
If there is another time, it will be 10. When the number of curves for y = 9, that is, the number of lookup tables is n, and the number of curves for y = 10 (the number of lookup tables) is m, the expected value of the output data y is x = 6. Is. Since this value should be 9.3, Since the total number of lookup tables LUT _{1 to} LUT _N is N, m + n = N ∴m = N−n (3) Substituting equation (3) into equation (2), n · 9 + (N− n) · 10 = 9.3N ∴n = 0.7N m = 0.3N That is, when x = 6, the number of lookup tables that output y = 9 is 0.7N, and the lookup that outputs y = 10. If the number M of up tables is 0.3N,
The output data y is 9.3 on average, which matches the target value.

When generalized, as shown in FIG. 2, the input data x
The target value y _j of the converted data corresponding to _j (j is a continuous integer) is an integer part I _j and a decimal part α _j (y _j = I _j + α _j) ;
If 0 ≦ α _j <1), the conversion data of (1-α _j ) · N lookup tables are set to y _j = I _j , and
The individual input / output conversion characteristics of the _N look-up tables LUT _{1 to} LUT N are determined under the rule that the conversion data of the remaining N look-up tables are set to y _j = I _j +1. .

However, since (1-α _j ) · N is not necessarily an integer, in that case, the number of lookup tables (1-α _j ) · N to be converted to y _j = I _j is rounded up, rounded down, or rounded up. Te (1-α _j) · to the nearest integer number _{n j} to _n, the number of look-up table for converting the y j _{= I} j +1 and _{m j} = _n-n j.

The lookup table LUT ₁ obtained in this way
A schematic diagram of input / output conversion characteristics of ˜LUT _N is shown in FIG.
The larger the value of the fractional part α _j, the larger the number of lookup tables to be converted into I _j +1. The smaller the value of α _j , I
The number of lookup tables converted to _j increases.

Incidentally, in FIG. 3, among the conversion data I _j and I _j +1,
_{I j} is concentrated in the look-up table LUT ₁ side, but _I j +1 are concentrated in a look-up table LUT _N side, without any need to limited to this, look-up tables _LUT 1 _I in ~LUT _N j +1 The distribution of and is completely arbitrary.

Further, when the target value _{y j} for the input data _{x j} is the fractional part alpha _j = 0 an integer, in all of the look-up table _LUT 1 _{~LUT N,} converted data _{y j} to the input data _{x j} is the _{I j} Become.

That is, for example, in Figure 7, since the target value y ₀ is 0 to the input data x _0, the converted data by the lookup table are all 0, and the target value y of the conversion data to the input data x _{15 Since 15} is 15, the conversion data by each lookup table is all 15. Similarly, when x = 8, y = 11.0.
That is, when x = 8, the output y has no quantization error and is output correctly. In this case, all of the plurality of lookup tables LUT _{1 to} LUT _N take the value of y = 11, so that 11 data are output accurately rather than randomly. On the other hand, in the conventional examples shown in FIGS. 8A and 8B, the output is (11 + noise), and an unnecessary error is added.

In the third view assumes the same characteristics as 0 all converted data I ₀ is at input x _0, all converted data I _N in the input x _N is a required value.

The statistical characteristics of the lookup tables LUT _{1 to} LUT _N having the input / output conversion characteristics as described above are smooth target characteristics like the curve A in FIG. 7. On the other hand, in the conventional method shown in FIG. 8A or 8B, the statistical characteristic of the look-up table remains the curve B. Note that the statistical characteristics are obtained by averaging output data y having random values when attention is paid to certain input data x.

As can be seen from the above comparison, a plurality of types of lookup tables LUT _{1 to} LU having different input / output conversion characteristics from each other.
By selecting T _N by the selection signal S _i consisting of random number data, it is possible to introduce randomness into the digital gradation conversion circuit without mixing noise.

Second Embodiment The gradation conversion circuit 11 of the second embodiment, as shown in FIG.
It has a single memory 12 and a random number generation circuit 13. N look-up table LUTs in this single memory 12
_{1 to} LUT _N are formed. As memory 12, R
It may be OM or RAM.

The selection signal S _i output from the random number generation circuit 13 is stored in the memory 12
Input to the upper address _AU of the input data x
Input to 12 lower addresses A _L. The capacity of the memory 12 is N times the capacity of one look-up table in the case of the first embodiment, but due to the recent increase in the capacity of memory devices, such a circuit does not increase in cost.

As an example, if the number of bits of the data x and y is 8 and the number of lookup tables LUT _{1 to} LUT _N is 16, the memory capacity is 2 ⁸ × 16 = 4K bytes, which is a general memory capacity of one chip. Value.

The method of creating the input / output conversion characteristics of the lookup tables LUT _{1 to} LUT _N is the same as in the first embodiment.
An example of a schematic diagram of input / output conversion characteristics is shown in FIG.

The combined signal of the selection signal S _i and the input data x _j is stored in the memory 12
When it is input as an address signal for
Pieces of any one of the look-up table LUT _i of the look-up table _LUT 1 _{~LUT N} is selected, input data corresponding to _{x j} of the table is output from the memory 12 as the output data _{y j.}

In each of the above embodiments, the lookup table LUT is used.
_{Although 1 to} LUT _N are formed in the ROM or RAM, the LS including the random number generation circuit may be used instead of the LS.
It goes without saying that it may be configured by I.

<Effects of the Invention> According to the present invention, since the above-described operation is performed, the average value (= expected value) of the output data with respect to the input data coincides with the target value, so that data corresponding to the required conversion characteristic is obtained. The problem of tone jump between adjacent pixel groups and false contours is solved, and subtle gradation changes are well reproduced.

Further, since the look-up table having plural kinds of conversion characteristics is randomly selected by the selection signal from the random number generation circuit, it is not necessary to increase the number of bits more than necessary and noise is not mixed.

[Brief description of drawings]

1 to 4 relate to a first embodiment of the present invention.
FIG. 2 is a block diagram of a gradation conversion circuit using a lookup table, FIG. 2 is a diagram showing the relationship between input data and target values,
FIG. 3 is a schematic diagram showing input / output conversion characteristics of a look-up table in the gradation conversion circuit, and FIG. 4 is a schematic block diagram of a color scanner in which the gradation conversion circuit is used. 5 and 6 relate to the second embodiment, FIG. 5 is a block diagram of a gradation conversion circuit, and FIG. 6 is a schematic diagram showing input / output conversion characteristics of a look-up table in the gradation conversion circuit. . FIG. 7 is an input / output conversion characteristic diagram of a general look-up table, and FIGS. 8A and 8B are block diagrams showing a conventional method. x, x _j …… Input data y, y _j …… Output data (converted data) _j …… Average value (= expected value) I _j …… Integer part α _j …… Decimal part S _i …… Selection signal LUT ₁ ~ LUT _N ... Look-up table 10, 13 ... Random number generation circuit 7, 11 ... Gradation conversion circuit

Claims (1)

[Claims] 1. N lookup tables (N is an integer of 2 or more) for outputting conversion data stored in advance when digital image input data and selection signals are input, and the N lookup tables. A random number generating circuit for randomly outputting the selection signal for selecting any one of the tables, and converting the target value y _j of the conversion data corresponding to the input data x _j (j is a continuous integer) into an integer. The part I _j and the decimal part α _j (y _j = I
_j + α _j ; 0 ≦ α _j <1), (1-α _j ) · N or the closest integer number of conversion data of the lookup tables are set to y _j = I _j , and N out of N Of the remaining lookup table conversion data of y _j = I _j +1
The gradation conversion circuit using the look-up table is characterized in that the individual input / output conversion characteristics of the N look-up tables are defined based on the above rule.