JP3039654B2 - Image reduction apparatus and method - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image reduction apparatus and method for reducing an image by forming reduced pixels with reference to a target pixel and surrounding pixels.

[Conventional technology]

Conventionally, a projection method or the like is known as an image reduction method for obtaining a smoothed image from an original size image and subsampling the pixels of the smoothed image to obtain a reduced image. The projection method is configured as shown in FIG.

Reference numeral 1 denotes a full-size image, which is generally a multivalued image. Reference numeral 2 denotes a low-pass filter, 3 denotes a sub-sampling unit, and 4 denotes an obtained reduced image. The low-pass filtering unit 2 performs low-pass filtering on the input full-size image depending on the reduced size, and obtains smoothed pixels. The sub-sampling unit 3 obtains the number of pixels required to form the reduced image 4 among the smoothed pixels. When the pixels of the input image are binary, as shown in FIG. 10, a binarizing section 5 is added after the sub-sampling section 3 in FIG. 9 to convert the multi-valued data calculated by the low-pass filter. Binarization is performed again to obtain a binary reduced image.

Now, for the sake of explanation, a case in which the image size is reduced to half the horizontal size is taken as an example. FIG. 11 shows a reference image of the low-pass filter in the original size image surrounded by a thick frame centering on the pixel 10f. Pixel 10f, 10a~10c, 10e, 10g, and x ₀ ~x ₈ as shown in FIG values of pixels of 10I～10k. Output X ₀ of the low-pass filter is expressed as using x ₀ ~x ₈ (1) formula.

By sub-sampling the result of the mark in FIG. 11 among the output results of the low-pass filter, a 1/2 reduced image can be obtained.

[Problems to be solved by the invention]

However, in the above-described conventional example, when the image is a character image or the like containing many thin lines, the thin lines in the reduced image are separated or unreproduced, and the quality of the image is remarkably deteriorated. For ease of explanation, a binary image is taken as an example. A thin slanted line as shown in FIG. 12 (a) is obtained by using the low-pass filter of the reference image of FIG.
Not reproduced at all. A thin line as shown in FIG. 12 (b) is reproduced if the pixel position to be sub-sampled is on the fine line, but is not reproduced at all if it is shifted, as in FIG. 12 (a).

[Means for solving the problem]

The present invention has been made in view of the above points, and in a configuration for generating a reduced image in which the original image is reduced in half in each of the vertical direction and the horizontal direction, disappearance of thin lines in the reduced image and thickening of the thin lines More specifically, the present invention is directed to an image reduction apparatus that generates a reduced image in which an original image is reduced by a factor of two in each of a vertical direction and a horizontal direction. A smoothing unit that performs a low-pass filtering process on each value of a pixel of interest in the image and a plurality of pixels around the pixel of interest, and a reduced pixel that forms the reduced image is formed by sub-sampling the output of the smoothing unit. Storing means for storing at least a reduced pixel formed immediately before and a reduced pixel at the same position in a previous line, which is a reduced pixel constituting the reduced pixel formed by the forming means. A step, values of a pixel of interest in the original image to be subjected to the low-pass filter processing by the smoothing unit and values of a plurality of pixels around the pixel of interest, and a reduced pixel formed at least immediately before stored in the storage unit Determining means for determining whether to correct the value of the reduced pixel formed by the forming means based on each value of the reduced pixel at the same position in the previous line, and correcting the value of the reduced pixel by the determining means Correction means for correcting the output value of the smoothing means in accordance with the judgment that the original image should be output. 1 / for each
A method for generating a reduced image reduced to 2, wherein a smoothing step of performing a low-pass filter process on each value of a pixel of interest in the original image and a plurality of pixels around the pixel of interest, and the smoothing step By sub-sampling the output obtained by the above, a determination step of determining the value of the reduced pixel constituting the reduced image, and a pixel of interest in the original image to be subjected to low-pass filter processing in the smoothing step and its Whether or not to correct the value of the reduced pixel determined in the determination step based on each value of the surrounding multiple images and at least each value of the reduced pixel formed immediately before and the reduced pixel at the same position on the previous line A determination step of determining whether or not the value of the reduced pixel in the determination step should be corrected,
And a correction step of correcting the output value obtained by the smoothing step.

FIG. 1 is a functional block diagram showing a configuration of a reference example of an image reduction device.

1 is a full size image to be input, 2 is a low-pass filter, 3
Is a sub-sampling unit, 4 is an obtained reduced image, 6 is a correction value calculator that obtains structural information from a low-pass filter reference pixel of the input original size image, determines whether or not there is correction, and calculates a correction value. It is an adder.

It is assumed that the value of the pixel of the original image 1 to be input is represented by 8 bits from 0 to 255, and the size to be reduced is 1/2. The input full-size image 1 is input to the low-pass filter 2 in units of three lines, and a smoothed image is obtained by a 3 × 3 matrix operation according to the above equation (1).

The original image 1 is input to the correction value calculator 6. FIG. 2 is a block diagram of the correction value calculator 6.

Reference numerals 20a to 20c denote terminals for inputting reference pixels of the row filter 2 of the original image 1 for each line. Reference numerals 21a to 21l denote latches for latching input pixel values and shifting data to the right according to timing. 22 and 26 are adders for adding nine inputs, and 23 is a divider for dividing the input by the number of reference pixels. 32 is a latch for latching data.
Reference numeral 24 denotes a subtractor that inputs nine subtracted numbers, subtracts separately input numbers from each of the subtracted numbers, and outputs the result. Reference numeral 25 denotes a squarer that outputs a squared number for each of the nine inputs.
A comparator 27 outputs the sign of the input having the largest absolute value among the nine inputs. Reference numeral 29 denotes an input terminal, and reference numeral 30 denotes a comparator for comparing two inputs to determine the magnitude. 28 is 2
A selector for inputting three select signals and switching and outputting three inputs, and 31 is an output terminal.

The pixel data input from terminals 20a to 20c is latch 21a.
The adder 22 obtains the sum of the pixel values of the nine pixels at the time when they are stored in .about.21i and outputs the sum to the divider 23. The divider 23 divides the sum of the pixel values by the total number of pixels to obtain an average value. At the next timing, the reference pixel block is shifted to the latches 21d to 21l, and the average value is latched to the latch 32. The subtractor 24 calculates an error between each pixel value of the latches 21d to 21l and the average value of the latch 32, and outputs the error to the squarer 25 and the comparator 27.

The comparator 27 detects the maximum error among the nine errors that have been input, and outputs a sign 1 bit of the maximum error (positive is represented by "0" and negative is represented by "1"). The squarer 25 squares each of the nine inputs to obtain a square error, and the adder 26 calculates the total square error. The comparator 30 compares the value input from the terminal 29, for example, 100,000, with the sum of the square errors, and outputs "1" if the sum is larger, and outputs "0" if the sum is smaller.

The selector 28 switches between the three inputs and outputs. These three inputs are 0, +100, -100. When the output of the comparator 30 is “1”, the correction is performed, and when the output is “0”, the correction is not performed.
When performing the correction, the comparator 27 performs a negative correction when the sign of the comparator 27 is negative, that is, “1”, and performs a positive correction when the sign of the comparator 27 is positive, that is, “0”.
For example, when no correction is performed, the output of the selector 28 is 0, when the negative correction is performed, the output is -100, and when the positive correction is performed, the output is +100. These results are the first from terminal 31
It is input to the adder 7 in the figure.

The adder 7 adds the output of the low-pass filter 2 and the correction value from the correction value calculator 6 to obtain the value of the reduced pixel.
The sub-sampling unit 3 obtains a reduced image 4 that is reduced by half in the horizontal direction.

As described above, in the configuration shown in FIG. 1, the structure information in the low-pass filter reference pixel block is extracted, and a thin line or the like is detected.
Since the output of the low-pass filter is corrected, a reduced image can be obtained by saving a thin line of a character or the like that disappears or is lost only with the low-pass filter.

〔Example〕

FIG. 3 is a functional block diagram of an embodiment of the image reducing apparatus according to the present invention. Reference numeral 101 denotes an original size image, reference numeral 102 denotes a low-pass filter, reference numeral 106 denotes a correction value calculator which determines whether or not there is correction, and calculates a correction value. Reference numeral 107 denotes an adder. 103 is a sub-sampling unit, and 105 is a binarization unit. Reference numeral 109 denotes a buffer for storing a part of the reduced image, and reference numeral 104 denotes the obtained reduced image.

Now, it is assumed that the pixel value of the input full-size image 101 is a binary value of 0 (white) or 1 (black), and the size to be reduced is 1/2.
It will be described as. The input full-size image 101 is input to the low-pass filter 102 in units of three lines, and a smoothed image is obtained by the above-described equation (1). The original image is also input to the correction value calculator 106 at the same time.

Here, the buffer 109 temporarily stores the already-reduced reduced pixels of the previous raster and the raster being processed among the already reduced images. Of these reduced pixels, the fourth
As shown in the drawing, the reduced image obtained immediately before the reduced pixel to be obtained and the value of the reduced pixel at the same position on the previous raster are also input to the correction value calculator 106. In FIG. 4, hatched pixels 110a to 110x are already obtained reduced images, and 111a is a reduced image to be obtained from now.

At this time, the correction value calculator 106 receives the reduced pixels 110o and 110x. FIG. 5 is a block diagram of the correction value calculator 106. Reference numeral 40 denotes a terminal for inputting reference image information of the low-pass filter, and reference numeral 41 denotes an input terminal for inputting information of peripheral reduced pixels from the buffer 109. 42 is ROM (Read Only Memory)
Is a LUT (Look Up Table). 43 is a selector for switching and outputting one of the three inputs, and 44 is an output terminal.

The reference image (9 pixels) input from the input terminal 40 and the surrounding reduced pixels (2 pixels) input from the input terminal 41 are the LUT 42
Is entered as the address. The LUT 42 determines whether or not to perform correction based on the input information of 11 bits. When performing correction, 2-bit information indicating whether the correction is negative or positive is stored in advance.

For example, if the reference pixels as FIG. 6 (a) is input, the X ₀ when obtaining the reduced pixel value X ₀ in accordance with the equation (1) 0.
If the threshold value T is 0.5, the reduced pixel becomes white. However, when the continuity of the line segment is considered, the reduced pixel must be black because the importance of information storage in the reduced image is emphasized. Similarly, when a reference pixel as shown in FIG. 6B is input, the reduced image must be black in consideration of the continuity of the line segment. However, in the case of FIG. 6C, when the sub-sampled circle is set as the center pixel of the reference pixel block, the result of the projection method is black in the left reference pixel block, and the reference pixel block in the right side. Then, it becomes black for the above-mentioned reason. In this, 2 of the full scale image
Since the problem that the line segment of the pixel width becomes two pixels even in the reduced image occurs, no correction is made to the projection method for the reference pixel block on the right side. In other words, even if the structure of the reference pixel block is exactly the same, there are some which are not corrected depending on the situation of surrounding pixels. Sixth
In the reference pixel block as shown in FIG. 7B, if the immediately preceding reduced pixel position is white, positive correction is performed, and if it is black, no correction is performed. Such information is determined and stored in the LUT 42 in advance. The meaning of the two bits of the output of the LUT 42 is as shown in Table 1.

In accordance with the output of the LUT 42, the selector 43 switches between three inputs and performs output. There are three inputs: 0, +0.5, and -0.5625. 0 is output when the input is not corrected, +0.5 when a positive correction is performed, and -0.5 when a negative correction is performed.
625 is selected and output from the terminal 44 to the adder 107 in FIG. The adder 107 performs addition of the output of the low-pass filter and the correction value, obtains the value of the reduced pixel, and
The number of pixels is reduced by half in the horizontal direction by 103, input to the binarization unit 105, and compared with the threshold value T = 0.5 to obtain a reduced binary image 104. This result is also output to the buffer 109 and fed back to the correction value calculator 106.

As described above, in the configuration of FIG. 3 according to the present invention, by using the structure information in the low-pass filter reference pixel block and the information of the reduced pixels already obtained, a thin line that disappears only with the low-pass filter is stored and the reference pixel is saved. It is possible to prevent thickening of a thin line, which is caused by the determination of only the structural information in the block.

FIG. 7 is a functional block diagram showing the configuration of another reference example of the image reduction device.

201 is a full-scale image to be input, 202 is a low-pass filter,
208 is a latch for timing adjustment, 206 is a correction value calculator for judging the presence or absence of correction and calculating a correction value, and 207 is an adder. 203 is a sub-sampling unit, and 205 is a binarizing unit. Reference numeral 204 denotes an obtained reduced image, and reference numeral 209 denotes a buffer for storing a part of the reduced image.

Now, it is assumed that the pixel value of the input full-size image 201 is a binary value of 0 (white) or 1 (black), and the size to be reduced is 1/2.
It will be described as. The input full-size image 201 is input to the low-pass filter 202 in units of three lines, and the image value of the smoothed image is obtained by the above-described equation (1) and latched on the latch 208. At the same time, the surrounding reduced pixels 110o and 110x as shown in FIG. 4 are input to the correction value calculator 206 from the output result of the low-pass filter, the reference pixel block, and the reduced pixels already reduced by the buffer 209.

FIG. 8 is a block diagram of the correction value calculator 206. 240 is a terminal for inputting reference pixel information of a low-pass filter, 241
Is an input terminal for inputting information on peripheral reduced pixels from the buffer 209. Reference numeral 242 denotes an LUT including a ROM. 243 is an input terminal for inputting the result of the low-pass filter, 244 is a binarizer for binarizing at a threshold T = 0.5, 245 is an exclusive OR circuit, 246 and 247 are 2-input and 1-output selectors, and 248 is Output terminal.

The reference pixel (9 pixels) input from the input 240 and the surrounding reduced pixel (2 pixels) input from the input terminal 241 are the LUT 242
Is entered as the address. The LUT 242 derives the binarized result without correction from the input 11-bit information when no correction is required in each case, and the binarized result after correction when correction is required. The result determined in advance as described above is stored. If FIG. 6 is taken as an example, in the case of FIG.
That is, black is output. In the case of FIG. 6B, when the result of the reduced pixel on the left side is 1 (black), the result is 0 (white), and the result is 0.
In the case of (white), 1 (black) is output.

The output of the low-pass filter is input from the input terminal 243, and binarized by the binarizer 244. The exclusive OR circuit 245 checks when the output of the binarizer 244 and the output of the LUT 242 are different, that is, whether or not the correction is necessary. When the output of the exclusive OR 245 is 0, the correction is not performed, and when the output is 1, the correction is performed. Selector 246 is 2
This is an input 1 output selector. When the result of the LUT 242 is 0, a negative correction value, that is, -0.3125 is output. When the result is 1, a positive correction value, that is, +0.25 is output. These correction values are determined as follows.

In the block represented by the reference pixel block in FIG. 6B, the output of the low-pass filter is the smallest among the reference pixel blocks to be positively corrected (0.25), and the threshold T = 0.
If the difference value from 5 is the largest, the positive correction value is 0.2
It becomes 5. Conversely, the negative image shown in FIG. 6 (b) also needs to be negatively corrected, and the low-pass filter output is the largest among the reference image blocks to be negatively corrected (0.7%).
5), assuming that the difference value from the threshold value T = 0.5 is the largest, the negative correction value is −0.3125 since the corrected value is smaller than 0.5.
Becomes

The selector 247 determines the correction value determined in this way,
Exclusive OR circuit for output 0 of correction value without correction
Switched by the output of 245, the adder of FIG.
Output to 207.

The adder 207 adds the output of the low-pass filter and the correction value to obtain the value of the reduced pixel,
The number of pixels is reduced by 1/2 in the horizontal direction,
The binary image 204 that is input to the value conversion unit 205 and is compared with the threshold value T = 0.5 is obtained. This result is also output to the buffer 209, and is used as information for determining whether or not subsequent pixel correction is necessary.

As described above, the configuration shown in FIG. 7 uses the structure information in the low-pass filter reference pixel block and the information of the already obtained reduced pixels, and further uses the output of the low-pass filter to save a thin line that disappears only with the low-pass filter. In addition, it is possible to prevent thickening of a thin line caused by judging only the structural information in the reference pixel block, and to control the image quality and the density by changing the weight coefficient of the low-pass filter and the threshold value for binarization. it can.

[Other Examples]

The sub-sampling method in each configuration described above is not limited to this, and it is also possible to select the center pixel position of the low-pass filter and perform it simultaneously with the low-pass filter. The threshold value for each correction value and binarization is not limited to this.
The shape and weighting factor of the low-pass filter are not limited to these. The nature and reduction ratio of the original image are not limited to these. Further, the position of the peripheral reduced pixel to be referred to is not limited to this.

As described above, the output value of the low-pass filter is corrected based on the structure information of the reference pixel of the low-pass filter and the output result of the low-pass filter, if necessary, and the information of the already obtained reduced pixel. By preventing thin lines from disappearing and disappearing, and by reproducing the thin lines as they are, information can be well preserved and image quality can be improved. There is also an effect that the density can be adjusted while maintaining the image quality by changing the binarization threshold. Even when the reduction is repeated several times to obtain a smaller reduced image, the information can be stored better.

[Effects of the Invention] As described above, according to the present invention, in a configuration for generating a reduced image in which the original image is reduced by half in both the vertical and horizontal directions, the target pixel in the original image and its surroundings A low-pass filter process is performed on each value of the complex pixel of, and a reduced pixel constituting a reduced image is formed by sub-sampling the smoothed output,
Based on each value of the target pixel in the original image to be subjected to the low-pass filter processing and a plurality of pixels around the target pixel, and at least each value of the reduced pixel formed immediately before and the reduced pixel at the same position in the previous line, It is determined whether or not the value of the reduced pixel to be formed is to be corrected, and the smoothed output value is corrected according to the determination that the value of the reduced pixel is to be corrected. Since the value of the reduced pixel constituting the reduced image reduced by half with respect to each can be corrected to a value in consideration of at least each value of the reduced pixel formed immediately before and the reduced pixel at the same position on the previous line. It is possible to prevent a thin line from disappearing or becoming thicker in a reduced image in a configuration in which a value of a reduced pixel forming a reduced image is formed by referring to each value of a target pixel in the original image and a plurality of pixels around the target pixel. Possible .

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a reference example of an image reduction device, FIG. 2 is a block diagram of a correction value calculator in FIG. 1, and FIG. 3 shows functions of an embodiment of the image reduction device according to the present invention. FIG. 4 is a block diagram of the surrounding reduced pixels, FIG. 5 is a block diagram of the correction value calculator in FIG. 3, and FIG. 6 is an example of a reference pixel block to be corrected. FIG. 7, FIG. 7 is a block diagram showing the configuration of another embodiment of the image reduction apparatus, FIG. 8 is a block diagram of the correction value calculator in FIG. 7, and FIGS. FIG. 11 is a diagram showing a matrix of a low-pass filter when the reduction ratio is 1/2, and FIG. 12 is a diagram showing a problem of the conventional example. 1,101,201 are full-size images, 2,102,202 are low-pass filters,
3,103,203 are sub-sampling units, 4,104,204 are reduced images, 5,105,205 are binarization units, 6,106,206 are correction value calculation units,
7,107,207 are adders, and 109,209 are buffers.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Akihiro Katayama 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Inventor Tadashi Yoshida 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon (56) References JP-A-63-219267 (JP, A)

Claims (2)

(57) [Claims] 1. An image processing method according to claim 1, further comprising:
An image reducing apparatus that generates a reduced image reduced to half, comprising: a smoothing unit that performs a low-pass filter process on each value of a target pixel in the original image and a plurality of pixels around the target pixel; Forming means for forming reduced pixels forming the reduced image by sub-sampling the output of the means; and reducing pixels forming the reduced image formed by the forming means, wherein the reduced pixels are formed at least immediately before A storage unit for storing the reduced pixel and the reduced pixel at the same position in the previous line; and each value of the target pixel in the original image to be subjected to the low-pass filter processing by the smoothing unit and a plurality of pixels around the target pixel, and Based on the values of at least the immediately preceding reduced pixel and the reduced pixel at the same position in the previous line stored in the storage unit, the form by the forming unit is used. Determining means for determining whether the value of the reduced pixel to be formed should be corrected, and correcting the output value of the smoothing means in accordance with the determination by the determining means that the value of the reduced pixel should be corrected. An image reduction device comprising: 2. The method according to claim 1, wherein the original image is divided into a vertical direction and a horizontal direction.
An image reduction method for generating a reduced image reduced to half, wherein a smoothing step of performing a low-pass filter process on each value of a target pixel in the original image and a plurality of pixels around the target pixel, Determining the value of the reduced pixel constituting the reduced image by sub-sampling the output obtained in the step; and a pixel of interest in the original image to be subjected to low-pass filtering in the smoothing step. Based on each value of a plurality of pixels around it, and at least each value of a reduced pixel formed immediately before and a reduced pixel at the same position on the previous line,
A determining step of determining whether the value of the reduced pixel determined in the determining step should be corrected; and a smoothing step in accordance with the determination that the value of the reduced pixel should be corrected in the determining step. A correction step of correcting the output value.