JPH043148B2 - - Google Patents

Description

[Detailed description of the invention]

The present invention relates to an improvement in a dither processing device used for dithering an image in an image processing device such as a facsimile machine or a digital copying machine. Dither processing is often used when expressing a halftone image using binary values of white and black. A device that performs this dither processing usually processes each pixel (i, j) of the input image.
The density level S (i, j) of the corresponding element T _D (k,
1) is used as a threshold and binarized into a white level (“0”) or a black level (“1”). That is, S(i,j)≧
If T _D (k, l), the output pixel is set to "1"; otherwise, the output pixel is set to "0". Here, (i,
The conversion between j) and (k,l) generally follows the following equation. k=Mod (i, M) ...Equation (1) (however, when k=0, let k=M) l=Mod (j, N) ...Equation (2) (however, when l=0 There are various types of dither matrices used for dither processing, and among them, the Bayer type and the dot concentration type are famous. However, no matter which dither matrix is used, if the threshold elements are fixed, moiré (periodic striped patterns) may occur in the output image depending on the nature of the input image, resulting in a significant deterioration in image quality. Sometimes. This is because the same threshold value element is used periodically and repeatedly, and when the input image has a periodic pattern such as a halftone dot print image, moire is likely to occur. In order to prevent this moiré, for example,
There is a method of randomly changing the threshold value of a dither matrix, as disclosed in Patent Publication No. 24634. According to such a method, moiré is certainly reduced. However, on the other hand, the distribution of threshold values tends to be uneven within a narrow pixel area, so when an image without moiré (a non-halftone image such as a photograph) is processed, grainy noise may occur in the output image. Image quality deteriorates. As described above, it is difficult to obtain high-quality output images for all types of input images using one type of dither matrix, and it is necessary to use different types of dither matrices depending on the nature of the input image. SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a dither processing device that automatically selects the type of dither matrix used for dither processing so as to obtain an output image of good quality. According to the inventor's research, the degree of psychological image quality deterioration of the output image f(x, y) after dither processing is F(ω _x ,
It can be expressed as the ratio of low frequency components of ω _y ).
Here, F (ω _x , ω _y ) is the power spectrum of f (x, y), which is obtained as the Fourier transform of the autocorrelation function of f (x, y), and therefore the proportion of low frequency components of the power spectrum. (Psychological image quality deterioration degree)
D _e is defined by the following equation. De＝∬ ^A/2 _A/2 F (ω _x , ω _y ) dω _x dω _y /P _dc ...Equation (3) However, ω _x and ω _y are the spatial frequencies in the main and sub-scanning directions of the image, and A is a parameter representing a low frequency boundary, and normally, if the period of the dither matrix is T, it can be set as A=π/T. Also P _dc
is the line spectrum of the DC component. Therefore, the dither processing device according to the present invention has a means for storing a plurality of types of dither matrix groups, a means for inputting an output image to obtain its autocorrelation function, and a power spectrum of the output image from the obtained autocorrelation function. means for calculating a proportion of low frequency components (psychological image quality deterioration degree); and means for appropriately reading out the stored dither matrix group and selecting a dither matrix that minimizes the psychological image quality deterioration degree. It is characterized by having the following. Hereinafter, one embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram of a dither processing apparatus according to the present invention. 100 is a comparison circuit and pixel transfer timing
The input image signal S _i is input pixel by pixel in synchronization with CLK1. One threshold element of one dither matrix read out from the table circuit 102 through the table reading circuit 101 is also input to the comparison circuit 100 . The comparison circuit 100 binarizes the input image signal S _i using this threshold element as a threshold value, and outputs it as an output image signal S ₀ . The table readout circuit 101 receives the pixel transfer timing CLK1 on each scanning line of the input image from a control circuit (not shown), the start timing CLK2 of each scanning line, and the selection signal SEL from the dither matrix selection circuit 106. be done. The table reading circuit 101 reads one dither matrix designated by the selection signal SEL from among the plurality of types of dither matrix groups stored in the table circuit 102.
The threshold elements corresponding to the pixels inputted to are sequentially read out and supplied to the comparison circuit 100. The correspondence relationship between this threshold element and the input pixel is expressed by the above equation (1) and (2).
Follow the formula. A plurality of types of dither matrices are stored in the table circuit 102. In this embodiment, a Bayer type fixed matrix of size M=N=4 as shown in FIG. Random dither matrices and conditional random dither matrices (both have a size of M=N=4) generated for each four pixel area are sequentially stored in the table circuit 102. Here, the random dither matrix mentioned above is obtained by randomly rearranging the threshold values from 1 to 15 over the entire dither matrix, for example, as shown in a, b, and c in Fig. 3 over time. It changes like that. The above conditional random dither matrix, for example, can be created by dividing the fixed dither matrix shown in Figure 2 into 4 parts vertically and horizontally for every 4 elements (as shown by the thick lines in Figure 2). This is obtained by randomly rearranging the threshold values, and an example thereof is shown in FIG. Details of these random dither matrices and conditional random dither matrices and their generation methods are disclosed in the specification attached to Japanese Patent Application No. 56-6150. A description of the specific configuration will be omitted. In this embodiment, the input image signal S _i changes the density level of each pixel of the input image from level 0 (white) to level
This is a signal obtained by quantizing to 16 values of 15 (black). 104 is a product-sum circuit, and 105 is a Fourier transform circuit. The product-sum circuit 104 is a circuit that calculates the autocorrelation function of the output image from the output image signal S ₀ . The Fourier transform circuit 105 calculates the above-mentioned autocorrelation function from this autocorrelation function.
The proportion De (psychological image quality deterioration degree) of low frequency components in the power spectrum of the output image is calculated according to equation (3), and the calculated value or a signal proportional to it is sent to the dither matrix selection circuit 106. do. The dither matrix selection circuit 106 receives a selection signal
The output signal value of the Fourier transform circuit 105 is checked while changing the type of dither table read out by the table reading circuit 101 as appropriate using SEL, and the value (that is, the degree of psychological deterioration of the output image) is the minimum value. Find the matrix and select signal
SEL specifies that the table reading circuit 101 read out the selected dither matrix. Such an operation of selecting the optimum dither matrix may be repeated at regular time intervals, or may be repeated after one or several scanning lines.
Alternatively, the dither matrix selection may be performed for the first few scan lines of the input image, and the selected optimal dither matrix may be specified intermittently until the final scan line of the input image. The former two methods are advantageous when processing an input image in which patterns with different properties coexist, and the latter method can be applied when processing an input image in which the pattern properties are uniform throughout. An example of the dither matrix selection circuit 106 described above is shown in FIG. 5 and will be described. In this example, a dither matrix to be used in the first three scan lines of an input image is selected, and the dither matrix is selected intermittently from the fourth scan line onwards. A counter 50 counts the number of scanning lines from the first scanning line to the third scanning line of the input image. The control circuit 51 sends the value of the counter 50 to the selector 52 and the table reading circuit 101 while the value is between 1 and 3. Therefore, in the first scan line, the first dither matrix is read out from the table circuit 102 (FIG. 1), in the second scan line, the second dither matrix is read out, and the third dither matrix is read out from the table circuit 102 (FIG. 1). In the third scan line, the third dither matrix is read out (in this example,
It is assumed that three types of dither matrices exist in the table circuit 102). On the other hand, the selector 52 sets the output of the Fourier transform circuit 105 for the first scanning line in the first stage ₅₃₁ of the register 53 in accordance with the count value sent from the control circuit 51. Similarly, 2
The outputs of the Fourier transform circuit 105 for the main and third scanning lines are set in the second stage 53 ₂ and third stage 53 ₃ of the register 53, respectively. The comparator 54 is connected to each stage 53 ₁ to 53 ₃ of the register 53.
, and outputs to the control circuit 51 the number of the register stage containing the minimum value, that is, the number of the dither matrix. The control circuit 51 continuously specifies the dither matrix number output from the comparator 54 to the table circuit 101 after the fourth scanning line. In this way, the dither processing device according to the present invention
By numerically evaluating the quality of the output image, the dither matrix that provides the highest quality of the output image is automatically selected and used. Therefore, it is possible to obtain a high-quality output image for input images with various properties and input images containing a mixture of patterns with different properties. Incidentally, for an input image with a uniform density level (level 7), the Bayer-type fixed dither matrix shown in FIG. 2, the random dither matrix shown in FIG. 3, and the conditions shown in FIG. 4 are used. When dithering is performed using each of the random dither matrices _with
It will look like Table 1. As mentioned above, the value of D _e corresponds to the degree of psychological image quality deterioration of the output image,
In this example, it can be seen that the image quality is best when using the Bayer type fixed dither matrix. Although numerical examples are not shown, for input images (such as halftone printed images) where moiré occurs with a fixed dither matrix,
The Bayer-type fixed dither matrix has the worst image quality, and better image quality can be obtained by using a random dither matrix or a conditional random dither matrix.

[Table] Although one embodiment has been described in detail, the present invention is not limited thereto, and various modifications may be made without departing from the gist of the present invention. For example, the types of dither matrices stored in the table circuit are not limited to those described above, and four or more types may be stored.
Product-sum circuits, Fourier transform circuits, dither matrix selection circuits, and even dither matrix generation circuits can of course be pure hardware circuits, but they can also be constructed in combination with software technology such as microprogramming technology. good. As described above, the present invention can provide an excellent dither processing device that can produce output images of high quality, and its effects are remarkable.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of a dither processing device according to the present invention, FIG. 2 is a diagram showing an example of a Bayer fixed dither matrix, and FIGS. 3 a to c are examples of a random dither matrix. A diagram showing
FIG. 4 is a diagram showing an example of a conditional random dither matrix, and FIG. 5 is a diagram showing a specific example of the dither matrix selection circuit in FIG. 1. 100... Comparison circuit, 101... Table reading circuit, 102... Table circuit, 103... Dither matrix generation circuit, 104... Product-sum circuit, 1
05... Fourier transform circuit, 106... dither matrix selection circuit.

Claims (1)

[Claims] 1. In a dither processing device that binarizes the density level of each pixel of an input image using the corresponding element of the dither matrix as a threshold and converts the input image into a binary output image, multiple types of dither matrix groups are stored. means for inputting an output image and determining its autocorrelation function; and means for calculating the proportion of low frequency components of the power spectrum of the output image (hereinafter referred to as psychological image quality deterioration degree) from the determined autocorrelation function. and means for appropriately reading out the stored dither matrix group and selecting a dither matrix that minimizes the degree of psychological image quality deterioration.