JPS6031383B2 - Image processing method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention particularly relates to an image processing method for restoring an image including halftones by binary output.

In general, in image input/output devices such as digital copying machines and facsimiles, when transmitting read image signals from the input side to the output side, the transmission line capacity and transmission time are reduced. A major challenge is how to reproduce an image that is close to the original image on the output side with a binary output while trying to improve the transmission efficiency of image signals such as digitization.

Conventionally, as a method for reproducing a high-gradation image by sampling an image containing half-tones such as a photograph pixel by pixel and processing the image signal obtained by performing multi-value quantization,
The difference between the density level of the pixel of interest (pixel to be processed) obtained by scanning the image on the input side and the density level of the pixel immediately before it is encoded and transmitted to the output side, and the density level is recorded on the output side. The differentially encoded signal is restored and multivalued or binary
A method of restoring a high gradation image using value output is being explored. However, in such conventional image restoration methods, a high gradation density level of at least 1 Naga tone is required to restore an image containing halftones such as a photograph with high image quality, and therefore the quantized image information is It becomes more than a Rousuge tone,
Even if the differential encoding is performed on the input side, the average bit length per pixel will exceed 2 bits/pixel (for example, in an image quantized with 1 tsubo total tone, the density level of the pixel of interest and the pixel before it) In many cases, the difference between the density level and the density level of The disadvantage is that it takes a long time.

The present invention has been made with these points in mind, and it quantizes images at low gradations pixel by pixel in order to optimize the encoding and transmission of image information. The present invention provides an image processing method capable of restoring an image according to the tone.

The image processing method according to the present invention takes advantage of the fact that, in images containing halftones, there is a very high density correlation between the pixel of interest and its surrounding pixels, and processes the pixel of interest quantized at low gradations. Using a specific model formula from the density level information in a specific pixel area centered on the pixel, the predicted density level of the pixel of interest, which has been increased in gradation, is calculated, and the calculated predicted density level is set in advance. By comparing it with a threshold, it is possible to perform a binary determination of the black level or white level of the pixel of interest.

Hereinafter, one embodiment of the present invention will be described in detail with reference to the accompanying drawings.

Figure 1 shows a 3x3 image area in which the image is divided into pixels, each pixel Pi-,,j-, ~P...,
FIG. 2 shows the density level information of each pixel when quantizing each pixel at a low gradation.

In the image input/output device described above, each of the density level information Qi-, ,j-, ~Qi-, , , shown in Fig. 2 is encoded and the encoded signal is transmitted to the output side. In this case, the quantization number on the input side is a low gradation, so for example, for 4 gradations, the encoded signal is 2 bits/pixel, and for 8 gradations, it is an encoded signal of 3 bits/pixel, which can be transmitted at high speed. At the same time, the encoding can be easily performed on the output side. In the pixel processing method according to the present invention,
Based on the image information decoded on the output side, the density level information Qi,j obtained by quantizing the pixel of interest Pi,j at a low gradation and the surrounding eight pixels Pi-
,,j-,~Pi-・,i,Pi+・,i〜PM,j+
, each concentration level information Qi-1, i-, ~Qi-1 small Q
By i ten,, i to Qi ten,, j+, the pixel of interest Pi
, j is calculated according to the following equation.

IM=k(a・Qi-,, j-, 10b・Qi, i-・10c・Qi+M-・10d・Qi-・, i+e・Qw+f・
QM, i+g・Qi-,,i ten, tenh・Qi,i ten, teni・Qi ten,,M)...'1
) Here, k is a modulation coefficient given by the number of output gradation levels/the number of input gradation levels, and is an integer of 2 or more,
In addition, a to i are weighting coefficients corresponding to each pixel, and a
+b+c+d+e+f+g+h+i=1.

Next, in the image processing method according to the present invention, the modulation coefficient k' of the equation '1-' is divided in advance into a matrix,
A threshold pattern T1,m in which a threshold corresponding to each pixel is set is prepared, and the threshold pattern corresponding to the predicted density level li,j of the pixel of interest obtained by formula m is compared and judged as follows. The black level and white level of the pixel signal to be output are determined.

(i} When li,j3T1,m, black level output thigh 1
When M3T1,m, the pixel of interest is processed above the white level output, from the first pixel P, , to the final pixel Pmax,m
By carrying out the process up to ax, it becomes possible to predictably restore a high-gradation, high-quality image from the image signal obtained by quantizing the low-gradation.

As a specific example, consider the case where the density of each pixel obtained by scanning an image is quantized in four gradations, and a binary image with a density level of 1-rougei is predictively restored from each density level information. , will be explained below. FIG. 3 shows an example of image information having four gradation density levels obtained by scanning and sampling an image. Such density level information obtained on the input side is encoded as is, converted into a coded string for one line, and transmitted to the output side. FIG. 4 shows the encoded density level information of the first column in the image of FIG. The signal encoded in this way is decoded on the output side so that it returns to the original state shown in Figure 3, but this is easy because each encoded density level information is 2 bits/pixel. This means that it can be decrypted to . Now, to simplify the explanation, the pixel of interest P:,j and its surrounding pixels include four pixels directly adjacent to it in the upper, lower, left, and right directions P;,
The predicted density level li,j is determined using each density level information of j-, , Pi-, , j, Pi+, , j, Pi, j+.

In this case, the modulation coefficient k is k = 16/4 = 4, and if each weighting coefficient b, d, e, f, h is a constant Q, then b + d + e + f + h = 1 to Q = 0.2, and the above [1} Formula becomes as follows.

IN=4(〇.2XQi,j-・10.2XQto,j
+〇． 2×Qi,i+0.2×Qi×,,i×0.2x
Qi, M) ...[2} Also, FIG. 5 shows the preset threshold pattern in this example, and shows the predicted density level li,j obtained by the above formula (■) and the corresponding threshold pattern. By making a comparison judgment with the threshold T1,m,
As described above, the black or white level of the pixel signal to be output is determined. FIG. 6 shows an output image predictively restored based on the image in FIG. 3 in this manner. in this way,
With the image processing method according to the present invention, a high-gradation output image can be obtained from low-gradation input image information at high speed and with high image quality.

FIG. 7 shows an example of a configuration for concretely implementing the image processing method according to the present invention. On the image input side A, the density information detected for each pixel by scanning the image is processed into a quantizer. 1 to perform low gradation quantization, each quantized signal is encoded by an encoder, and the encoded signal is transmitted to the output side. On the output side B, the encoded signal sent from the input side A is decoded by the decoder 3, and its contents are temporarily stored collectively in the memory 4 to store the density level information of each pixel in the specific pixel area. Based on the stored contents, the arithmetic unit 5 performs the calculation according to equation (1), and the predicted density level li,j of the pixel of interest based on the calculation result and the threshold read from the ROM 6 in which the threshold pattern is stored in advance. T1,m are compared by a comparator 7, and a pixel signal of black or white level is outputted to the outside according to the result of the judgment of magnitude.In addition, 8 in the figure indicates a pixel signal given from the outside. An address indicator is shown for specifying an address for recalling pixel contents necessary for an operation into the storage device 4 in response to an address signal sent thereto, and for specifying an address for recalling a corresponding threshold and value from the ROM 6. As described above, in the image processing method according to the present invention, two images including halftones are processed.
When performing predictive restoration using value output, the density information is read by quantizing the image at a low gradation for each pixel, and the density level of the read pixel of interest and a specific pixel area around the pixel of interest are calculated. The predicted density level of the pixel of interest is calculated from the density level of each pixel in the pixel using a specific calculation formula that takes into account the modulation coefficient and the weighting coefficient of each pixel, and the predicted density level of the pixel of interest is calculated from the density level of each pixel within the preset threshold pattern. The system compares the threshold value corresponding to the pixel and determines the black or white level depending on the result of determining whether the pixel is large or small. It has the excellent advantage that when transmitting the encoded signal to the output side, it can be processed efficiently and particularly at high speed, and moreover, a high-gradation image can be restored with high quality on the output side.

[Brief explanation of the drawing]

Figure 1 is a diagram showing a specific pixel area where the image is sampled pixel by pixel, Figure 2 is a diagram showing the density level information of each pixel in low gradations, and Figure 3 is a diagram showing the image in four gradations. FIG. 4 is a diagram showing an encoded signal obtained by encoding the first column, FIG. 5 is a diagram showing an example of a threshold pattern, and FIG. FIG. 7 is a block diagram showing a configuration example for concretely implementing the image processing method according to the present invention. 1...Quantizer, 2...Encoder, 3...
...Nacoda, 4...Memory unit, 5...Arithmetic unit, 6
...ROM, 7...Comparator, 8...Address indicator. Figure 4 Figure 1 Figure 2 Figure 3 Figure 5 Figure 6 Figure 7

Claims (1)

[Claims] 1 In a method of restoring an image containing halftones using a binary image, a modulation coefficient is calculated from the density level of each pixel in a specific pixel area centered on the pixel of interest, based on image information read at low gradation levels for each pixel. Then, the predicted density level of the pixel of interest is determined using a specific model formula that takes into account the weight of each pixel, and the predicted density level of the pixel of interest is calculated based on the calculated density level and the threshold pattern set in advance according to the modulation coefficient. An image processing method characterized in that a pixel of interest is binarized by comparison with a corresponding storage field.