JPH1132206A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processor which can suppress the occurrence of voiding or color image deterioration at the time of compositing images and can image enhance compression efficiency at the time of performing separation and compression. SOLUTION: In an image processor, an image separating section 2 separates an image inputted to an image inputting section 1 into an image S indicating the state of a character or a line drawing, an image P indicating the color information of the character or line drawing, and an image I which is obtained by deleting the data of characters and of line drawings from the inputted image. An image-storing section 3 temporarily stores an image I in which the existing area of the characters or line drawings is voided. In the image element in the voided area of the image I, an image processing section 4 fills up the voided area with an image element value by referring to the image element value in the periphery of the voided area. Thus, even the existing areas of the characters and line drawings in the image also become flat image, and the characters and line drawings can be composited with each other without causing voids, and compression efficiency can be enhanced.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for separating input image information into a plurality of pieces of image information and performing processing.

[0002]

2. Description of the Related Art When an input image is stored or transferred, it is desired to reduce the data amount as much as possible. In order to reduce the amount of data, an input image may be separated into, for example, character or line drawing information and one or more pieces of other image information, and optimal data compression processing may be performed on each of the information. Also,
Before the data compression processing, the image may be converted into a different resolution for each separated image information to further reduce the data amount.

Japanese Patent Laying-Open No. 4-105178 discloses a configuration in which a character area, a chart area, and an image area are separated and extracted from an input image, converted into data suitable for information included in each area, and then reconstructed. .

FIG. 16 is an explanatory diagram of a specific example of image separation processing in a conventional image processing apparatus. FIG. 16A shows an input image, and the hatched portion is an image, and a character “ABC” is superimposed on the image. When the image separation processing is performed on such an input image, the image is separated into an image representing a character area as shown in FIG. 16B and an image representing an image area as shown in FIG. In the above document, character recognition processing is performed on the separated character area to reduce the data amount.

Generally, the image portion has a large data amount. For example, when an A4 size image is read from a 600 dpi scanner with 8 bits for each color of RGB, the data amount is about 108 Mbytes. At present, generally, the resolution of characters and line drawings is required to be 600 dpi or higher, but the resolution of image portions may be lower than that of characters and line drawings. For this reason, the resolution of the image portion is lower than that of the character portion, or the amount of data is reduced by using a compression method typified by the JPEG baseline method, and data storage and communication are performed.

For example, consider a case where an image shown in FIG. 16C, which is obtained by picking up only an image portion from the input image shown in FIG. At the time of output, the image held at the low resolution is converted into the original resolution and combined with the character part image shown in FIG. At this time, when the image portion is converted to the original resolution, FIG.
There may be differences between the image shown in FIG.
White spots may occur at the boundary between the character portion and the image portion.

On the other hand, as shown in FIG. 16 (C), on an image obtained by separating only the image area, the area where the character was present becomes blank. In particular, when many fine characters are written on an image, white spots frequently appear on an image in which only the image area is separated, and high-frequency components in the image become large. When such an image is compressed using a variable-length and irreversible data compression method using a discrete cosine transform such as a JPEG baseline method, there is also a problem that the compression rate is reduced and the image quality after compression / decompression is deteriorated. is there.

[0008]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-mentioned circumstances, and has been made in consideration of the above circumstances. It is intended to provide a device.

[0009]

SUMMARY OF THE INVENTION The present invention separates input image information into first image information consisting of character or line drawing information and second image information consisting of information other than character or line drawing information. Regarding the separated second image information,
The image information is embedded in a part or all of the area separated as the first image information with reference to pixels around the area. As a result, even if the second image information is converted to a low resolution, restored to the original resolution, and combined with the first image information, no white spot is caused by the embedded image information. Further, even if a character or a line drawing exists in the image portion, the second image information can embed the image information by referring to surrounding pixels in the portion of the character or the line drawing so as to increase the compression efficiency, for example. . Therefore, the compression efficiency of the second image information does not decrease in the portion where the character or the line image exists, and the compression can be performed efficiently, and the image quality at the time of restoration can be maintained.

[0010]

FIG. 1 is a block diagram showing an embodiment of an image processing apparatus according to the present invention. In the drawing, 1 is an image input unit, 2 is an image separation unit, 3, 5 to 7 are image storage units, 4
Denotes an image processing unit. The image input unit 1 is configured by an image input device such as an image scanner or an interface unit for inputting an image from a network, and passes the input image to the image separation unit 2. The input image is
It may be a color image or a grayscale image.

The image separating section 2 separates an input image into a character or line drawing area and other areas. In this example, the character or line drawing area is separated into two data, "data indicating the shape of the character or line drawing" and "palette data indicating the color information of the character or line drawing". The image in the area other than the character or line drawing area is “data obtained by deleting character or line drawing data” from the input image. Hereinafter, “data indicating the shape of a character or a line drawing” is abbreviated as an image S, “palette data indicating color information of a character or a line drawing” is abbreviated as an image P, and “data obtained by deleting character or line drawing data” is abbreviated as an image I. The separated image I is stored in the image storage unit 3
, The image S is stored in the image storage unit 6, and the image P is stored in the image storage unit 7, respectively. As a method of separating an image in the image separating section 2, many methods have been proposed, such as a method described in Japanese Patent Application Laid-Open No. 3-126180 and a method described in Japanese Patent Application Laid-Open No. 4-105178 described above. It is optional to adopt a suitable method.

The image processing section 4 uses the image S stored in the image storage section 6 for a portion extracted as a character or a line drawing in the image I stored in the image storage section 3 and missing. A correction process is performed, and the result is stored in the image storage unit 5.

The image storage unit 5 stores the image I corrected by the image processing unit 4. The corrected image I, the image S, and the image P stored in the image storage units 5, 6, and 7
Are output respectively. Each output image can be converted in resolution or encoded, stored in a database, or communicated via a network or the like. At the time of resolution conversion or encoding, each image may be converted into a resolution according to the characteristics of each image, or may be encoded using an encoding method according to the characteristics of each image. As an example, the image I can be encoded and stored by a JPEG baseline method or the like with a reduced resolution. The image S can be encoded and stored at the same resolution, for example, by the JBIG method. The image P can be stored with a very low resolution, for example, encoded by the JPEG baseline method or the like.

Hereinafter, an example of the processing operation in the image processing section 4 will be described. FIG. 2 is an explanatory diagram of image data. In the following description, the image is treated as two-dimensional coordinates, and the upper left coordinates of the image are represented by (i, j) = (1,
1), the upper right coordinate is (i, j) = (i _max , 1), the lower left coordinate is (i, j) = (1, j _max ), and the lower right coordinate is (i, j) = (i _max , j _max ).

In the following description, S (i, j) is a pixel value corresponding to the coordinates (i, j) of the image S, and I (i, j).
j) represents a pixel value corresponding to the coordinates (i, j) of the image I. Here, the image S is 1-bit data, and each pixel value S (i, j) = 1 indicates “character or line drawing”, and S (i, j) = 0 indicates that the pixel is “character”. Or line drawing ". The image I is data having a gradation, and in the case of color, the pixel value I is
(I, j) is, for example, "red (R), blue (B), green (G)"
Is represented by multi-valued data (for example, 8 bits and 256 gradations). In the following description, one color component is shown. However, when the color components are divided into a plurality of color components, the same operation may be performed on each color.

FIGS. 3 and 4 are flowcharts showing a first operation example of the image processing unit. Here, the image processing unit 4
Shows an example in which a missing portion extracted as a character or a line drawing in the image I is embedded using pixel values at both ends of the portion. More specifically, the image S is referred to for each line (each j), and a point at which the pixel value changes from 0 to 1 is detected and set as a left end (start point) of a portion where a character or a line drawing exists. Similarly, the point at which the pixel value of the image S changes from 1 to 0 is detected to determine the right end (end point). A portion from the detected start point to the detected end point is a blank portion in the image I. In this example, as a process for filling the blank portion,
The left side of the center is filled with the pixel value of the leftmost image I, and the right side of the center is filled with the pixel value of the rightmost image I. Thereby, it is extracted as a character or a line drawing in the image I, and the blank portion is filled for each line.

First, in step S21, a line j to be processed is initialized to 0, and in step S22, 1 is added to j to make the next line a processing line. Immediately after the start of the process, 1 is added to j = 0, and j = 1, indicating the first line. In S22, the column i to be processed in the processing line j is initialized to 0, and in S23, 1 is added to i, and the next pixel is set as the target pixel. At the beginning of each line,
1 is added to i = 0, and i = 1, indicating the first pixel of the line.

In S23 to S29, the starting point of the blank area in the image I is detected. After updating i in S23, S2
At 4, it is determined whether or not it is the start point by referring to the image S.
That is, the pixel value S (i, j) at the position corresponding to the target pixel in the image S is 0, and the next pixel value S (i + 1, j)
Is 1 or not. When this condition is satisfied, the pixel value I having the coordinate corresponding to the image I in the variable α is determined in S47 as the start point of the character or line drawing area.
(I, j) is stored, the coordinate i is stored in a variable A, and S
Go to 30. The variable α stores the pixel value at the start point of the blank area, and the variable A stores the coordinates of the start point.

If the condition of S24 is not satisfied, a blank area may be formed from the head of the line. Therefore, in S25, it is determined whether or not i = 1, that is, whether or not the head of the line. In the first case, it is determined in S28 whether the pixel value S (i, j) of the image S is 1 or not. When the pixel value S (i, j) of the image S is 1 at the head of the line, a blank area is formed from the head of the line.
In this case, 0 is stored in the variable α as the pixel value to be filled in S29, and 1 is stored in the variable A as the coordinates of the start point,
Proceed to S30.

If it is determined in step S25 that the pixel of interest is not at the head of the line, it is determined in step S26 whether or not the pixel of interest has reached one pixel before the right end of the line. The process proceeds to S39 to shift to the processing of the line. Since the next pixel does not exist for the rightmost pixel, it is not necessary to perform the determination in S24, and the process of detecting the start point is not performed. If the start point is not detected in S24 and the pixel of interest is two or more pixels at the right end of the line in S26 or it is determined that the start point of the line is in S25, but the start point is not a character or a line drawing in S28 In this case, the process returns to S23 and i
And shifts the pixel of interest to the next pixel.

In the process of repeating such processing,
If the start point of the blank portion on the line j (including the case starting from the head of the line) is detected, the process proceeds to S30. S30 to S34 are processing for detecting the end point of the blank area. After adding 1 to i in S30 and moving the target pixel to the next pixel, in S31, the pixel value S (i, j) at the position corresponding to the target pixel in the image S is 1, and
It is determined whether or not the next pixel value S (i + 1, j) is 0. If the condition is satisfied, the pixel value I (i + 1, j) having the corresponding coordinates in the image I is stored in the variable β in S33, and the coordinate i + 1 is substituted for the variable B, and the process proceeds to S35. The variable β stores the pixel value at the end point of the blank area, and the variable B stores the coordinates of the end point.

If the condition is not satisfied in S31, it is determined in S32 whether or not the target pixel is a pixel immediately before the right end. That is, it is determined that the blank area continues to the right end of the line. If the blank area has reached the right end of the line, 0 is stored as a pixel value in a variable β in S34, and i is set in a variable B as coordinates of an end point.
_{Stores max} . If the end point is not detected in S31 and the target pixel is two or more pixels at the right end of the line,
Returning to step 30, 1 is added to i to make the next pixel the target pixel.

By repeating such processing,
When the start point of the blank area is detected in the processing of S23 to S29, the end point of the blank area is detected. And S
Proceed to 35.

In steps S35 to S37, processing for embedding the pixel values stored in the blank area is performed. First, in S35, the pixel value (value of variable α) at the start point of the blank area is compared with the pixel value (value of variable β) at the end point.
If the values are the same, the pixel value I (A, A,
j) to I (B, j) are uniformly set to the value of the variable α or the variable β. If the values of the variable α and the variable β are different, in S36, the pixel values I (A, j) to I
(A + [(BA) / 2], j) is the value of the variable α, and pixel values I (A + [(BA) / 2] + 1, j) to I (B,
j) is the value of the variable β. Here, [] represents a Gaussian symbol. In this way, the blank area is filled with the pixel value of the start point or the end point.

In S38, it is determined whether or not the pixel of interest has reached the right end of the line, that is, whether or not the processing of one line has been completed. If it is in the middle of the line, the flow returns to S23 to start the blank area. Continue finding points. When the processing of one line is completed, it is determined whether or not the processing line is the last line in S39, and when it is not the last line, the process returns to S22, and the processing is continued with the next line as the processing line. The processing is repeated up to the last line, and when the processing of the last line ends, the processing in the image processing unit 4 ends.

In the above example, when the pixel values at the start point and the end point of the blank area are different, the center point C = A + [(B
-A) / 2], the pixel value at the start point is on the left side, and the pixel value at the end point is on the right side. However, the present invention is not limited to this, and the boundary can be arbitrarily determined as long as it depends on the start point and the end point. Of course, the entire white area may be filled with either the pixel value of the start point or the pixel value of the end point.

In the above-described first operation example, the processing is performed for each line, so that the required memory amount may be small. Further, the processing time can be reduced because the pixel values of continuous white spots are converted at once, instead of one pixel at a time.
Moreover, in the processed image I, since the same pixel value continues from the start point and the end point in the portion where the character or the line drawing was present, for example, a compression method for detecting a change point for each line is used. When the image I is encoded, encoding efficiency can be improved.

The above-described first operation example will be described using a specific example. FIG. 5 is an explanatory diagram of a specific example of the input image and each separated image. FIG. 5A shows a specific example of an input image, in which “ABC” is displayed on a halftone image such as a photograph.
This is an image with characters written on it. The halftone portion is hatched according to the density. Such an input image is separated by the image separation unit 2 into an image I shown in FIG. 5B, an image S shown in FIG. 5C, and an image P shown in FIG. In this state, the character "A"
The portion where "BC" is written is in a blank state as shown in FIG.

FIG. 6 is an explanatory diagram of a specific example of pixel values in each of the separated images and the processed image I. FIG.
FIGS. 6B and 6A show part of the pixel values on the horizontal line 11 in FIGS. 6B and 5C. FIG. 6 shows a case of a very low resolution to simplify the description. In the image I shown in FIG. 6B, the pixel value is 8 bits, "0" is white, "255" is black, and the hatched portion in FIG. The resulting portion is referred to as a pixel value 50, and the boundary is referred to as a pixel value 90. As a result of the separation by the image separation unit 2, the character "AB" in the image I shown in FIG.
The portion where "C" is deleted has a pixel value of 0 as shown in FIG. In the image S separated as a character or a line drawing, as shown in FIG. 6A, the value is 1 only in a portion where the character "ABC" exists. In this example, the density and the like of the character “ABC” are separated into the image P.

Now, it is assumed that the processing line j is the horizontal line 11 shown in FIGS. 5B and 5C. According to the flowcharts shown in FIGS. 3 and 4, the process proceeds while shifting the target pixel to the right in order from the left end of this line. When i = p, S from FIG.
Since (p, j) = 0 and S (p + 1, j) = 1, S2
Condition 4 is satisfied, and the starting point in the image I is detected. In S27, I (p, j) = 20 is stored in the variable α, and p is stored in the variable A.

Next, when i = q, S (q, j) = 1, S
(Q + 1, j) = 0, and the condition of S31 is satisfied.
As a result, the end point of the blank portion of the image I is detected, and S
33, I (q + 1, j) = 20 and B
Is stored in q + 1.

Then, a blank portion from the position p stored in the variable A to the position q + 1 stored in the variable B is filled. In S35, when the values of the variable α and the variable β are compared, since the values are equal, the blank portion is filled with the value 20 of the variable α or the variable β in S37. As a result, FIG.
As shown in (C), the blank area from the position p to q + 1 has a value of 20, and the value 2 including the blank area is included.
The portion of 0 is continuous.

Similarly, the region from the position r to s + 1,
Each blank portion of the region from the position t to u + 1 and the region from the position v to w + 1 is filled with the pixel value 50 of the start point or the end point. Further, from position x to y
In the white area up to +1, the pixel value α at the start point is 50,
The pixel value β at the end point is 20, and α and β are not equal.
Therefore, in S36, the pixel value 50 for the positions x + 1 and x + 2 and the pixel value 20 for the positions y−1 and y
Is buried.

In this way, a series of pixel values as shown in FIG. 6C is obtained. This shows an image in which a blank portion existing in FIG. 6B is filled.
Further, as shown in FIG. 6 (C), by filling the blank portion, the pixel value is continuous around the blank portion, and the section having the same pixel value is longer. Thus, for example, MH
In the case of encoding using run length, such as encoding, or predictive encoding for each line, since the run length becomes longer and prediction is more likely to be achieved, it is possible to improve encoding efficiency. it can.

Further, as shown in FIG. 5B, the image I having a blank portion is converted into a low-resolution image in order to reduce the amount of data and stored, and is restored to the original resolution at the time of output. When it is returned, the edge of the blank portion may be jagged, for example, and may not return to the original shape. If the image S and the image P shown in FIGS. 5C and 5D are combined in such a state, a white spot may occur around the character. However, in the present invention, since the blank portion is filled, the boundary with the character does not exist in the image I and is not affected by the resolution conversion. Therefore, there is no possibility that white spots will occur in the image after composition,
It is possible to output a good image.

Further, in the first operation example, since the processing for each line is performed, the memory required is very small, and the processing is simple by setting the pixel value to be embedded to be the same as the peripheral pixel. Yes, high-speed processing is possible.

FIGS. 7 and 8 are flowcharts showing a second operation example of the image processing section. In the above-described first operation example, when the blank portion is filled, the pixel values of the start point and the end point of the blank portion are used as they are. In the second operation example, a case will be described in which a pixel value of a blank portion is newly created from pixel values of a start point and an end point of a blank portion.

In FIGS. 7 and 8, parts performing the same processing as in FIGS. 3 and 4 are denoted by the same reference numerals, and description thereof will be omitted. The difference from the first operation example described above is that
The only difference is that the process of S36 is changed to the process of S41. By the processing up to S35, the coordinates of the start point of the blank area and its pixel value, and the coordinates of the end point and its pixel value are obtained. In S35, the pixel value α at the start point
If the pixel value β is not equal to the pixel value β at the end point, a pixel value of a blank portion is generated in S41.

In S41, the pixel values of the blank portion, that is, the pixel values I (A, j) to I (B, j) are determined as follows according to the distance from the start point A and the end point B, respectively. . I (i, j) = {(i−A) β + (B−i) α} / (B−A) (1) Pixel values I (A, j) to I (B, j) are calculated as (1) After the conversion by the expression, the process proceeds to S38, and the process is continued until the process is completed.

FIG. 9 is an explanatory diagram of a specific example of the pixel values of the image I in the second operation example of the image processing section. The first mentioned above
5A, the input image shown in FIG.
It is assumed that the images are separated into the images shown in (B) to (D).
FIGS. 9 (A) and 9 (B) are the same as FIGS. 6 (A) and 6 (B), and some pixels on the horizontal line 11 of the images S and I shown in FIGS. 5 (C) and 5 (B). Specific examples of values are shown. This second
In the operation example of (1), the portion surrounded by the dashed line
This is a part that performs an operation different from the operation example of FIG.

The operation of a specific example will be described with reference to the flowcharts shown in FIGS. 7 and 8, but only the parts that are different from the first operation example will be described. Positions p to q
+1, positions r to s + 1, positions t to u + 1, positions v to w +
Since the pixel values at the start point and the end point of each white area are equal to each other, the white area is filled with the pixel values at the start point or the end point in S37.

A start point is detected at the position x, and the position is stored in the variable A, and the pixel value 50 is stored in the variable α. Also, an end point is detected at the position y,
The position y + 1 is stored in the variable B and the pixel value 2
0 is stored in the variable β. Since the value of the variable α is different from the value of the variable β in S35, a process of filling a blank area is performed in S41. For example, I (x + 1, j)
From the above equation (1), (1 · β + 4 · α) / 5 = (20 + 200) / 5 = 44, and the pixel value 44 is embedded. Similarly, position y
Up to the pixel. FIG. 9C shows the result of the pixel filling in this way. From position x to position y +
It can be seen that the pixel values up to 1 change stepwise.

As described above, by embedding the pixels in the white area in a stepwise manner, it is possible to reduce a rapid change in gradation and suppress generation of high frequency components. Therefore, for example, when using an encoding method that involves conversion to frequency components such as discrete cosine transform, it is possible to reduce the code amount and suppress the quality degradation of the decoded image.

Of course, since the portion of the image I which is separated as a character or a line drawing and becomes blank is buried, the image I is converted to a low resolution and then converted back to the original resolution to be converted to a character or a line drawing. Even when the images are combined, no white spots occur at the edges of the characters and line drawings, and a good combined image can be obtained. Also, in the second operation example, since processing is performed for each line, there is an advantage that a very small memory is required.

Next, a third operation example of the image processing section 4 will be described. In the first and second operation examples described above, the image I
In filling the white spot, the pixel values of the white spot are determined with reference to two surrounding pixels on the same line. In the third operation example, two peripheral pixels on the same column are referenced together with two peripheral pixels on the same line, and a total of four pixels are referred to.
An example is shown in which each pixel value of a white area is newly created from the pixel value of a pixel.

FIGS. 10 to 12 show a third example of the image processing unit.
6 is a flowchart showing an operation example of the above. S51 to S53
Is a process similar to S21 to S23 in FIG. 3. First, in S21, j is initialized to 0, and then in S22, 1 is added to j to make the next line a processing line. Also, in S22, i = 0 is initialized, and in S23, 1 is added to i, and the next pixel is set as a target pixel.

S53 to S59 correspond to S23 to S in FIG.
This is the same processing as in step 29, and detects the start point of the blank area in the image I. After updating i in S53, it is determined whether or not it is the start point by referring to the image S in S54. That is, the pixel value S at the position corresponding to the pixel of interest in the image S
It is detected whether (i, j) is 0 and the next pixel value S (i + 1, j) is 1. If this condition is satisfied, the variable α is determined in S57 as the start point of the character or line drawing area.
1 has a pixel value I (i, i,
j) is stored, and its coordinate i is stored in the variable A1. Further, the coordinate j at this time is stored in the variable C, and the process proceeds to S60. The variable α1 stores the pixel value at the start point in the line direction of the blank area, and the variable A1 stores the coordinates of the start point in the line direction.

If the condition of S54 is not satisfied, a blank area may be formed from the head of the line. Therefore, in S55, it is determined whether or not i = 1, that is, whether or not the head of the line. In the first case, it is determined in S58 whether the pixel value S (i, j) of the image S is 1 or not. When the pixel value S (i, j) of the image S is 1 at the head of the line, a blank area is formed from the head of the line.
In this case, in S59, 0 is stored in the variable α1 as the pixel value of the start point in the line direction, and 1 is stored in the variable A1 as the coordinates of the start point in the line direction. Further, the coordinate j at this time is stored in the variable C, and the process proceeds to S60.

If the target pixel is not at the head of the line in S55, it is determined in S56 whether the target pixel has reached one pixel before the right end of the line. The process proceeds to S82 in order to shift to the processing of the line. Since the next pixel does not exist for the rightmost pixel, it is not necessary to perform the determination in S54, and the process of detecting the start point is not performed. If the start point is not detected in S54 and the pixel of interest is two or more pixels at the right end of the line in S56, or it is determined in S55 that the start point is the line, but in S58 the start point is not a character or a line drawing. In this case, the process returns to S53, and i
And shifts the pixel of interest to the next pixel.

In the process of repeating such processing,
The start point of the blank portion on line j can be detected. When the start point is detected, the process proceeds to S60. S60-S64
Then, similarly to S30 to S34 in FIG. 3, the end point in the line direction of the blank area is detected. After adding 1 to i in S60 to shift the target pixel to the next pixel, in S61, the pixel value S at the position corresponding to the target pixel in the image S
It is determined whether (i, j) is 1 and the next pixel value S (i + 1, j) is 0. If the condition is satisfied, in step S63, the pixel value I (i
+1 and j), and substitutes the coordinate i + 1 for the variable B1, and proceeds to S65. The variable β1 stores the pixel value at the end point in the line direction of the white area, and the variable B1 stores the coordinates of the end point in the line direction.

If the condition is not satisfied in S61, it is determined in S62 whether or not the pixel of interest is a pixel immediately before the right end. That is, it is determined that the blank area continues to the right end of the line. If the blank area has reached the right end of the line, 0 is stored as a pixel value in a variable β1 and i _max is stored as a coordinate of an end point in a variable B1 in S64. If the end point is not detected in S61 and the target pixel is two or more pixels at the right end of the line, the process returns to S60, adds 1 to i, and sets the next pixel as the target pixel.

By repeating such processing,
When the start point of the white area in the line direction is detected in the processing of S53 to S59, the end point of the white area in the line direction can be detected. Then, the process proceeds to S65.

In steps S65 to S80, a process of calculating and embedding the respective pixel values from the pixel next to the start point of the blank area detected as described above to the pixel immediately before the end point is performed. In S65, i is set to a variable A1, which is the starting point in the line direction of the white area, and in S66, 1 is added to i to set a pixel in which a pixel value is to be embedded as the next pixel. For example, immediately after i is set to the variable A1 in S65, the pixel is the pixel next to the start point of the blank area. Then, pixel values are calculated and embedded in S67 to S79, and in S80, it is determined whether or not the coordinate i is the coordinate of the pixel immediately before the end point B1 in the line direction of the white area, that is, all the coordinates in the line direction of the white area. It is determined whether or not the embedding process has been completed for the pixel. If there is any unprocessed pixel, the process returns to S66, where 1 is added to the coordinate i, and the next pixel is set as the pixel to be embedded with the pixel value. Repeat the process.

In steps S67 to S79, a process of calculating and embedding the pixel value to be embedded is performed. In this operation example, the pixel values of the two pixels at the start point and the end point of the blank area on the same column are also referred to. For this purpose, in S67 to S71, the starting point of the blank area in the column direction is obtained, and in S72 to S71.
At 77, the end point of the blank area in the column direction is obtained.

In steps S67 to S71, the coordinates j in the column direction are reduced for the pixel in which the pixel value is to be embedded.
Find the starting point in the column direction. In S67, 1 is subtracted from j. In S68, the pixel value S (i, j) of the image S at the same column position of the line j is 0, and the pixel value S of the image S at the same column position of the next line j + 1.
It is determined whether (i, j + 1) is 1 or not. If the condition is satisfied, the pixel value I (i, j) having the corresponding coordinates in the image I is stored in the variable α2 in S70, and the coordinates j are substituted for the variable A2, and the process proceeds to S72. The variable α2 stores the pixel value at the start point in the column direction of the white area, and the variable A2 stores the coordinates of the start point in the column direction.

If the condition is not satisfied in S68, it is determined in S69 whether j = 1, that is, whether the line has reached the upper end line. In this determination, it is determined that the blank area is continuous up to the top line. If the blank area has reached the top line, S7
1 stores 0 as a pixel value in a variable α2,
1 is stored in 2 as the coordinates of the start point. If the end point is not detected in S68 and the line does not reach the upper end line, the process returns to S67, and 1 is subtracted from j to determine whether the same column position one line before is the start point in the column direction of the blank area in the column direction. Determine whether or not. By repeating such processing, it is possible to obtain the starting point in the column direction of the blank area for the pixel in which the pixel value in column i is to be embedded.

Similarly, in steps S72 to S77, the end point in the column direction of the blank area in column i is obtained. In S72, j is set to the line (C-1) before the original processing line, and 1 is added to j in S73. In S74, it is determined whether the pixel value S (i, j) of the image S on the line j is 1 and the pixel value S (i, j + 1) of the image S on the next line j + 1 is 0. If the condition is satisfied, the pixel value I (i, j + 1) having the corresponding coordinates in the image I is stored in the variable β2 in S76, and the coordinate j + 1 is substituted for the variable B2, and the process proceeds to S78. The variable β2 stores the pixel value at the end point in the column direction of the blank area, and the variable B2 stores the coordinates of the end point in the column direction.

If the condition is not satisfied in S74, it is determined in S75 whether the last line has been reached. That is, it is determined that the blank area continues to the last line. If the blank area reaches the last line, 0 is stored as a pixel value in a variable β2 and j _max is stored as a coordinate of an end point in a variable B2 in S77. If the end point is not detected in S74 and it is not the last line, the process returns to S73, and 1 is added to j to make a determination on a pixel in the same column of the next line. By repeating such processing, the end point of the blank area in column i can be detected.

Next, in step S78, the line j is returned to the coordinates stored in the variable C, and in step S79, the pixel value of the pixel I (i, j) in the blank area is calculated and embedded. The pixel value of the pixel I (i, j) can be obtained by the following equation (2). I (i, j) = {(B1-i + B2-A2) α1 + (i-A1 + B2-A2) β1 + (B1-A1 + B2-C) α2 + (B1-A1 + C-A2) β2} / 3 (B1-A1 + B2-A2) .. (2) This equation calculates a pixel value according to the distance from four points of the line direction start point A1, end point B1, column direction start point A2, and end point B2. Note that the pixel value I
The calculation formula for determining (i, j) is not limited to formula (2), and may be arbitrarily determined as long as the calculation formula depends on the values of the four surrounding pixels of the blank area and the distance from the target pixel to the four surrounding pixels. Can be.

In S80, it is determined whether or not i = B1-1, that is, whether or not the embedding processing has been completed up to the pixel immediately before the end point in the line direction of the blank area, and unprocessed pixels still remain. In this case, the flow returns to S66, and the same processing is repeated until i = B1-1. When i = B1-1, the process proceeds to S81.

In S81, it is determined whether or not the processing has been completed up to the right end of the processing line. If the processing has not been completed up to the right end, the process returns to S53 to detect another blank area, and the processing is repeated. When the processing is completed up to the rightmost pixel of the processing line, S8
In step 2, it is determined whether or not the processing has been completed up to the last line. If the processing has not reached the last line, the process returns to step S52 and the same processing is performed until all the processing on the last line is completed. When the processing of the last line ends, the processing of the image processing unit 4 ends.

FIG. 13 is an explanatory diagram of a specific example of the pixel values of the image I separated in the third operation example of the image processing unit.
4 is an explanatory diagram of a specific example of the image S, and FIG. 15 is an explanatory diagram of changes in pixel values of a part of the image I before and after the embedding process. In this example, 6 × 8
This shows a case where an image composed of pixels is separated, and the image P is omitted. For convenience, column numbers 1 to 6 and line numbers 1 to 8 are given. 13 and FIG.
In FIG. 7, the pixel indicated by '*' is not related to the description, and the description of the pixel value is omitted. In FIG. 15, only the portions of columns 2 to 5 of the line 3 surrounded by the dashed line in FIGS.
(A)) and the pixel value after the embedding process (FIG. 15 (B)).

The operation of the specific example will be described below with reference to the flowcharts shown in FIGS. Now, it is assumed that line 3 is a processing line. When i = 1 in S53, S (1,3) = 0, S in S54
(2, 3) = 1, and the starting point in the line direction of the blank area is detected. Therefore, in S57, α1 = I
(1,3) = 20, A1 = 1, and C = 3 are stored.

Next, when i = 5 in S60, S (5,3) = 1 and S (6,3) = 0 in S61, and the end point of the blank area in the line direction is detected. for that reason,
In S63, β1 = I (6,3) = 50 and B1 = 6 are stored.

In step S65, i = 1, and in step S66, i = 1 + 1.
= 2, j is reduced by 1 in S67 to 2; S68
In S, since S (2,2) = 0 and S (2,3) = 1, the starting point in the column direction is detected.
2 = I (2,2) = 20 and A2 = 2 are stored.

Subsequently, in S72, j = 3-1 = 2, and in S7
After adding 1 to 3 and setting j = 3, the condition of S74 is determined, but the condition is not satisfied. The addition of the value of j is repeated in S73, and when j = 7, S (2,7) = 1, S (2,
8) = 0, thereby satisfying the condition of S74. That is, the end point in the column direction was detected. In S76, β2 =
I (2,8) = 20 and B2 = 8 are stored.

In this way, the starting point A1 = 1 in the line direction of the blank area, the pixel value α1 = 20, the ending point B1 = 6 in the line direction, the pixel value β1 = 50 are obtained, and the white point is further obtained. For the pixel (2, 3) in the missing area, the starting point A2 = 2 in the column direction, the pixel value α2 = 20, the ending point B2 = 8 in the column direction, and the pixel value β2 = 20 were obtained. After returning to j = C = 3 in S78, when I (2,3) is calculated according to the above equation (2) in S79, I (2,3) = ｛(6−2 + 8−2) × 20 + (2 -1
+ 8-2) × 50 + (6-1 + 8-3) × 20 + (6-
1 + 3-2) × 20｝ / {3 × (6-1 + 8-2)}
26. This value is embedded as I (2, 3) as shown in FIG.

Since i has not reached the right end of the line 3 in S80, the flow returns to S66, and 1 is added to i to make i = 3.
When the start point and end point of the white area in the column direction in the column of i = 3 are similarly obtained, the start point A2 = 1 in the column direction, the pixel value α2 = 20, the end point B2 = 5 in the column direction, The pixel value β2 = 90 is obtained.
From equation (2), I (3,3) = 44 is obtained. Similarly, I (4,3) = 53 and I (5,3) = 44 are obtained, and the values of I (2,3) to I (5,3) are obtained as shown in FIG.
Is converted as shown in

In the above example, line 3 has been described, but the same embedding process is performed for other lines. In the image I converted in this way, pixels that are extracted as characters or line drawings and are in a white spot state are filled with values corresponding to changes in pixel values not only in the i-coordinate direction but also in the j-coordinate direction. I have. Therefore, the converted image I
In the document, a portion where a character or a line drawing was present is buried so as to smoothly change two-dimensionally, thereby preventing an increase in high frequency components. Therefore, even when encoding is performed by employing an encoding method that encodes two-dimensional frequency components such as two-dimensional discrete cosine transform, the amount of code is reduced, and the quality of a decoded image is reduced. Can be prevented.

Of course, since the portions which are separated as characters and line drawings and become blank are filled in, the image I is converted to a low resolution, then converted back to the original resolution and synthesized with the characters and line drawings. Also, no white spots occur at the edges of characters and line drawings, and a good composite image can be obtained. At this time,
Since the pixel value after conversion is determined by referring not only to the i-coordinate direction but also to the values of the nearest neighbor pixels in the j-coordinate direction,
When combining with the character portion, no problem occurs in the i-coordinate direction.

In the above description, the image input in the image separation unit 2 is separated into three images, namely, an image I, an image S, and an image P. For example, for a color image, for example, R,
What is necessary is just to separate into three images for each color axis corresponding to each color coordinate system such as G, B, C, M, and Y. Here, the image S is a binary image, and the image P is an image having density, in which characters and line drawings are separated. However, the present invention is not limited to this. The image S and the image P may be separated as one character / line drawing image having a density without being different images. In this case, when comparing the values with reference to the image S in the above description, it is sufficient to check whether the value is 0 or any other value. Further, it may be configured such that attribute data is generated when the input image is separated, and the generated attribute data is also stored.

[0072]

As is apparent from the above description, according to the present invention, the input image information is converted into first image information consisting of character or line drawing information and second image information consisting of information other than character or line drawing information. When the image information is separated into the second image information and the image information is deleted corresponding to the first image information in the second image information and becomes a white spot, the image information is referred to by referring to pixels surrounding the area. Embed As a result, even when the second image information is held at a low resolution, converted to the original resolution at the time of outputting the image, and combined with the first image information, the outline of the character written on the image has a white spot. And a good composite image can be obtained.

Further, even when many fine characters and line drawings are written on the image, white areas in the regions where the characters and line drawings existed in the second image information are eliminated, and pixel values as flat as possible are reduced. Since the image has an image, it is possible to reduce the amount of code at the time of encoding, and to suppress a decrease in image quality after decoding. For example, even when a variable length and irreversible data compression method using a discrete cosine transform such as a JPEG baseline method is used, it is possible to prevent a reduction in compression rate and a deterioration in image quality after decompression. There is.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating an embodiment of an image processing apparatus according to the present invention.

FIG. 2 is an explanatory diagram of image data.

FIG. 3 is a flowchart illustrating a first operation example of the image processing unit.

FIG. 4 is a flowchart (continued) illustrating a first operation example of the image processing unit.

FIG. 5 is an explanatory diagram of a specific example of an input image and separated images.

FIG. 6 is an explanatory diagram of a specific example of pixel values in each of the separated images and the processed image I.

FIG. 7 is a flowchart illustrating a second operation example of the image processing unit.

FIG. 8 is a flowchart (continued) illustrating a second operation example of the image processing unit.

FIG. 9 is an explanatory diagram of a specific example of pixel values of an image I in a second operation example of the image processing unit.

FIG. 10 is a flowchart illustrating a third operation example of the image processing unit.

FIG. 11 is a flowchart (continued) illustrating a third operation example of the image processing unit.

FIG. 12 is a flowchart (continued) illustrating a third operation example of the image processing unit.

FIG. 13 is an explanatory diagram of a specific example of pixel values of an image I separated in a third operation example of the image processing unit.

FIG. 14 is an explanatory diagram of a specific example of the image S separated in the third operation example of the image processing unit.

FIG. 15 is an explanatory diagram of changes in pixel values before and after embedding processing of a part of the image I separated in the third operation example of the image processing unit.

FIG. 16 is an explanatory diagram of a specific example of image separation processing in a conventional image processing apparatus.

[Explanation of symbols]

1 image input unit, 2 image separation unit, 3, 5 to 7 image storage unit, 4 image processing unit.

Claims (3)

[Claims] 1. Separating means for separating input image information into first image information consisting of character or line drawing information and second image information consisting of information other than character or line drawing information, and separating by the separating means. Image information embedding means for embedding image information in a part or all of the area separated as the first image information with respect to the second image information by referring to pixels around the area. Image processing device. 2. The image information embedding unit refers to the peripheral pixels located on the same line in the vertical direction or the horizontal direction in the target pixel in the area, and determines a pixel of the target pixel based on a pixel value of the peripheral pixel. The image processing apparatus according to claim 1, wherein the value is determined. 3. The image information embedding unit determines a pixel value of the target pixel based on a distance between the target pixel and the peripheral pixel with respect to the target pixel and a pixel value of the peripheral pixel. An image processing apparatus according to claim 1.