JPH10108011A - Data-processing unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To allow the processing unit to apply division of image data to such areas as character, photographic and background areas and to apply compensation to each of the data suitable for each area consistently with high efficiency. SOLUTION: Rectangular areas are detected, unified and the result is added to a rectangular list by retrieving black pixels from an image that has been reduced and binarized from an original image duing preliminary scanning. Pixel data of each rectangular area extracted, based on the coordinates in the rectangular list among pixel data as a result of main scanning area given to an image discrimination circuit 135, in which each image characteristic is discriminated and the pixel data are coded by a low-frequency image high efficiency coding circuit 152 or a high-frequency image high efficiency coding circuit 153, depending on the image characteristic discriminated. Pixel data other than the data in the rectangular areas are coded by a background high efficiency coding circuit 154. After the end of coding processing, data in the rectangular list are compressed and stored in a main body memory 160. When an image is output, the rectangular data are decoded, and the image is expanded based thereon.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a laser printer or a digital copier which compresses image data in which multi-valued pixel data of one pixel and character data which can be expressed in a binary form are mixed, writes the image data in an image memory, and reads out the image data. And the like.

[0002]

2. Description of the Related Art Conventionally, data processing apparatuses such as laser printers and digital copiers compress image data in which one pixel includes multi-valued pixel data and one pixel includes binary character data, and store the compressed image data in an image memory. Some are written and then decompressed and read.

[0003]

However, characters can be represented by one bit, but in image data, one pixel is represented by multiple values in order to maintain image quality. Therefore, if you want to unify image data in which image data and character data are mixed into a multi-valued data format and write it to the image memory,
Binary data such as characters wastes storage space, while
In order to unify image data in which image data and character data are mixed into a binary data format and write the image data into an image memory, it is necessary to reduce image quality by error diffusion or the like in order to store image data and the like in binary.

[0004] Image compression compresses the amount of data by removing the spatial and temporal redundancy of the image, and is divided into information source coding and entropy coding. Source coding is a process that uses the correlation between pixels and frames and spatial spectrum bias to remove redundant parts of the image, and entropy coding uses the statistical bias of the source-coded data. This is a process for removing redundant parts. JPEG compression and Lempel-
There is Ziv compression and the like, and suitable applications differ depending on the compression method.

For example, JPEG is a compression method suitable for a photographic image in which the density difference between adjacent pixels is small and has a high compression ratio, and Lempel-Ziv is a compression method suitable for an edge image such as a character. The compression ratio is low because of this method. Since the background area is expressed as data of a certain level, the compression efficiency can be made higher than that of characters and photographs, and it is desirable that reversible compression be possible. Therefore, if a unified compression method is adopted without distinguishing a background area that is neither a character nor a photograph from an image area such as a character or a photograph, appropriate compression cannot be performed for each character, photograph, and background.

In order to solve the above-mentioned problem, the image data may be divided into character, photograph, and background areas, and compression suitable for the characteristics of each area may be performed. A method and apparatus for doing so have not yet been proposed.

An object of the present invention is to solve the above technical problems,
It is an object of the present invention to provide a data processing device to which a high-quality and high-compression-efficiency compression method can be applied according to image characteristics.

[0008]

The above object is achieved by the following constitution.

(1) A scanner for reading multi-valued pixel data from an original image in which characters and photographic images are mixed, and reduced data indicating an average level of two-dimensionally adjacent multi-valued pixel data transmitted from the scanner. , A binarizing means for binarizing the reduced data obtained from the reducing processing means, and a rectangular area on a page determined from the binary data obtained from the binarizing means. Rectangular data generating means for generating rectangular data of a rectangular area; a rectangular data memory for storing rectangular data from the rectangular data generating means; and a page memory for storing multi-valued pixel data from the scanner in page units. Image determination means for detecting image characteristics from multi-valued pixel data read from the page memory included in a rectangular area defined by the rectangular data; The multi-valued pixel data read from the page memory based on the determination result of the image determining means is converted into high-efficiency coding means for low-frequency images, high-efficiency coding means for high-frequency images, and high-efficiency coding for background. Extraction processing means for sending the data to any one of the means, and the high-efficiency coding means for the low-frequency image, the high-efficiency coding means for the high-frequency image, and the high-efficiency coding means for the background. A data processing apparatus comprising: a storage control unit that stores data in a memory; and a plurality of expansion processing units that expand high-efficiency code data read from the main body memory into image data.

The reason for dividing the original image into rectangular areas in the present invention is as follows.

In general, newspapers, catalogs, reports, and the like, in which characters and photographic images are mixed, have recently been increasing in which characters and photographic images are mixed. In many cases, a character region is inserted in a rectangular shape for each paragraph, and a photographic image is also often inserted in a rectangular shape. Therefore, if the image data read from the scanner is divided into rectangular regions, there is a high possibility that the image data can be divided into a character region and a photographic image region. The rectangular data obtained by the region dividing process is constituted by coordinates of one vertex of the rectangle and data indicating the height and width of the rectangle.

The extraction processing means of the present invention can not only extract a character area and a photographic image from an original image but also extract only a background area from the original image. Since the pixel data can be handled as pixel data, the compression efficiency of image data can be improved.

(2) The data according to (1), wherein the rectangular data generation means uses a search element that dynamically changes a search range and performs an integration process so that the divided rectangular areas do not overlap. Processing equipment.

In the rectangular area search processing of the present invention, a rectangular area can be specified by obtaining one vertex and its diagonal vertex. Therefore, in order to simplify the search processing, the search element is performed only in the horizontal and vertical directions, and The search range of the searcher can be dynamically changed so that the search range of the searcher is increased at the beginning of the search and the search range of the searcher is reduced near the apex of the rectangular area. Therefore, the search processing of the present invention can accurately and quickly search near the vertices of the rectangular area. The searcher moves over a small gap such as between characters or character strings. This also enables high-speed processing.

(3) The data processing apparatus according to (1), wherein the image discriminating means detects an image characteristic from a difference value between adjacent pixel data.

(4) The high-efficiency encoding means for high-frequency image is JPEG compression suitable for an image having little change between adjacent pixels and having high correlation between pixels. Is a compression suitable for an edge image having a large change between adjacent pixels and a low correlation between the pixels, and the background high-efficiency encoding means encodes continuous constant-level pixel data with high efficiency. (1) The data processing device according to (1).

(5) The data processing apparatus according to (1), wherein the rectangular data is encoded with high efficiency and stored in a main body memory.

[0018]

FIG. 1 is a block diagram showing a data processing circuit according to the present embodiment.

The data processing apparatus according to the present embodiment is applied to an electrophotographic digital copying machine for reproducing a multi-valued image, and includes a scanner I / F 111, a scan buffer 112, a reduction processing circuit 121, Processing circuit 122
, A memory control circuit 123, a reduced image memory 124, a rectangular division processing circuit 131, an extraction processing circuit 132, a buffer memory control circuit 133, and a rectangular data buffer memory 1.
34, image discriminating circuit 135, and page memory control circuit 14
1, a page memory 142, a log conversion circuit 151, a high-efficiency encoding circuit 152 for a low-frequency image, a high-efficiency encoding circuit 153 for a high-frequency image, a high-efficiency encoding circuit 154 for the background, an encoded data storage control circuit 155, and a rectangle. Data compression circuit 1
56, body memory 160, decoding control circuit 171, low frequency image decoding circuit 172, high frequency image decoding circuit 1
73, a background decoding circuit 174, a developed image storage control circuit 175, and an output interface 180.

The scanner I / F 111 sends multi-valued pixel data from an original image in which characters and photographic images are mixed to the scan buffer 112 and the page memory control circuit 141.

The reduction processing circuit 121 includes a scanner I / F 1
11 to obtain a reduced image of the original image obtained from
An average gray level of pixel data of adjacent 256 gray levels is calculated by a secondary matrix of 10 × 10 pixels obtained by the prescan, and the pixel data of the average gray level is sent to the binarization processing circuit 122. is there. Therefore, the reduction processing circuit 121 is a circuit having a function of reducing the matrix area of 10 × 10 pixels of the original image to one pixel. The reason for reducing the matrix of 10 × 10 pixels to one pixel in the present embodiment is that one-tenth is obtained when a 400 dpi image is reproduced.
Since the resolution is 1.57 pixels / mm at a resolution of 40 dpi, 12th class characters (3 × 3 m
This is because the number of pixels in m) is about 2.4 pixels, and if a search is made for three or more pixels, it is possible to search over a gap of a character string from half a character to one character. Therefore, the matrix size is changed by the character series used for the original image. The reduction method employed in the reduction processing circuit 121 may be any method as long as the image information of the image area is accurately reflected.

The binarization processing circuit 122 is a circuit for sending out the reduced data inputted from the reduction processing circuit 121 at a predetermined threshold value to "0" and "1" to the memory control circuit 123. The binarization method employed in the binarization processing circuit 122 may be any method that can accurately extract a black pixel region, and employs an adaptive binarization method or a binarization method using a histogram.

The memory control circuit 123 is a binary processing circuit 1
This is a circuit which writes or reads the binary data from the memory 22 to the reduced image memory 124 and sends it to the rectangular division processing circuit 131.

The rectangular division processing circuit 131 corresponds to a rectangular data generating means for searching for a rectangular area on the original image and generating rectangular data of the rectangular area. Since a rectangular area is searched using binary data, high-speed processing is possible. In the present embodiment, the rectangular data is composed of the coordinates of the upper left vertex of the extracted rectangular area and data indicating the height and width of the rectangle. The rectangular data is not limited to this, but may be any data as long as it indicates any vertex coordinates of the rectangular area and the height and width of the rectangle.

The buffer memory control circuit 133 is a memory control circuit for writing or reading rectangular data or identification results in the rectangular data buffer memory 134.

The rectangular data buffer memory 134 stores the rectangular data and the result of determining the image characteristics.

The page memory 142 has a scanner I / F 11.
This is a memory for storing multi-valued pixel data input from 1 in page units, and is equivalent to the input / output memory in the present embodiment. Therefore, it is connected to a writing control circuit for performing an exposure process of an electrophotographic process. Output I / F1
80 connected.

The extraction processing circuit 132 divides the pixel data read from the page memory 142 into rectangular areas and non-rectangular areas based on the rectangular data, and sends them to the image discriminating circuit 135 or the background high-efficiency encoding circuit 154. is there. Here, the non-rectangular area is an area other than the character image area or the photographic image area, in other words, corresponds to the background area. Accordingly, the extraction processing circuit 132 can send continuous pixel data of a constant luminance level to the background high-efficiency encoding circuit 154. Pixel data belonging to the rectangular area is sent to the image discriminating circuit 135 continuously for each rectangular area.

The rectangular data buffer memory 134
Is stored in the reduction processing circuit 121.
Since it is multi-valued data of one pixel from 0 × 10 pixels, since it is necessary to match the coordinate position with the original image data,
Rectangle data must be multiplied by ten. In particular,
The upper left vertex (x ₁ , y ₁ ) of the rectangular data before expansion and its diagonal coordinates (x ₂ , y ₂ ), and the upper left vertex (x ′ ₁ , y ′ ₁ ) of the rectangular data after expansion and its diagonal Assuming that the coordinates are (x ′ ₂ , y ′ ₂ ), there is the following relationship. Assuming α = 10 and β = α / 2, x ′ ₁ = αx ₁ −β, y ′ ₁ = αy ₁ −β, x ′ ₂ = α
x ₂ + β-1, y ′ ₂ = αy ₂ + β-1.

The image discriminating circuit 135 detects an edge from a difference between adjacent pixels of multi-valued pixel data included in the rectangular area defined by the rectangular data, and outputs a character when the number of edges exceeds a set threshold value. The image characteristic of the area defined by the rectangular data is detected by discriminating the area as a photographic area if the area is not exceeded, and the result of the determination is added to the rectangular data and sent to the buffer memory control circuit 133. , The multi-valued pixel data belonging to the area defined by the rectangular data is selectively transmitted to the high-efficiency encoding circuit 152 for low-frequency images or the high-efficiency encoding circuit 153 for high-frequency images.

The high-efficiency encoding circuit 152 for low-frequency image
JPEG compression that can compress a photographic image with a small difference in density between adjacent pixels with high quality and high efficiency is adopted.

The outline of the JPEG compression method will be described below.
It is divided into blocks of 8 × 8 pixels as information source coding, and is subjected to two-dimensional DCT (discrete cosine transform) to convert to a 8 × 8 cosine transform matrix corresponding to a spatial frequency component. DCT coefficients are linearly quantized with a different step size for each coefficient using a quantization table. Low-frequency component coefficients that greatly affect image quality are finely quantized, and high-frequency component coefficients that have relatively little effect on image quality are quantized roughly. The value of the low-frequency component becomes large by the quantization process,
The value of the high frequency component is almost “0”.

Thereafter, as the entropy coding, the DC component and the AC component are separately converted into a binary sequence and Huffman coding is performed. Specifically, the coding of the DC component performs Huffman coding on the difference value from the DC component of the immediately preceding block,
The coding of the AC component is performed by performing a zigzag scan so that the coefficient of “0” is continuous from the low frequency component to the high frequency component and arranging the coefficient in a one-dimensional manner, and then a run length representing the length of the continuous coefficient of “0”. Then, the encoding is performed using a coefficient value other than “0” that follows. By changing the quantization level, it is possible to control the compression ratio and the image quality, which are in conflict with each other. Note that Huffman coding is lossless coding.

The high-efficiency encoding circuit 153 for high-frequency images
Lempel-Ziv suitable for compression of characters with many edges
Adopts compression. The Lempel-Ziv compression algorithm encodes an input character string into a code based on a conversion table, registers a character string not registered in the conversion table in the conversion table, and dynamically updates the conversion table in both the encoder and the decoder. Is what you do.

The background high-efficiency encoding circuit 154 is a circuit for compressing the background region into one-dimensional data in raster scan order because the background region can be treated as continuous data of a constant level.

The coded data storage control circuit 155 converts the coded data transmitted from the low-frequency image high-efficiency coding circuit 152, the high-frequency image high-efficiency coding circuit 153, and the background high-efficiency coding circuit 154. This is a write control circuit stored in the main body memory 160.

The rectangular data compression circuit 156 encodes the rectangular data and the determination result with high efficiency.

The main body memory 160 stores data that has been efficiently coded by a compression method selected for each image characteristic of a rectangular area of an arbitrary size, rectangular area data, and a discrimination code indicating the image characteristic of the rectangular area. is there.

The image expansion control circuit 171 reads out the high-efficiency coded data at the address calculated from the rectangular data, and decodes the low-frequency image based on the discrimination code added to the rectangular data. This is a circuit to be sent to one of the decoding circuit 173 and the background decoding circuit 174.

The low-frequency image decoding circuit 172 has a DCT
The decoding circuit is based on. The entropy decoding circuit, the Huffman coding table, the inverse quantization circuit, the quantization table and the IDCT circuit, decode the coded data into multi-valued pixel data and store the expanded image Control circuit 175
Circuit.

The high-frequency image decoding circuit 173 executes a Lempel-Ziv decompression algorithm for decoding into character data. The decoding circuit 173 decodes a high-frequency image into characters including edges with many high-frequency components while creating a conversion table. And decodes the data into multi-valued pixel data and sends it to the developed image storage control circuit 175.

The background decoding circuit 174 is a circuit for decoding into background pixel data and sending it to the developed image storage control circuit 175.

The expanded image storage control circuit 175 is a circuit for storing multi-valued pixel data in the page memory 142 in accordance with an address obtained based on the rectangular data read from the main body memory 160, and is input from the scanner I / F 111. Output I / F1 by substantially restoring the output image data
80.

The above is the schematic configuration of the data processing circuit according to the present embodiment.

Next, a schematic processing operation of the data processing circuit according to the present embodiment will be described.

When an image input command is issued, the scanner I /
From F <b> 111, multi-value pixel data by pre-scan is read into the scan buffer 112. Reduction processing circuit 1
Reference numeral 21 performs reduction processing in units of the capacity of the scan buffer 112, and the binarization processing circuit 122 binarizes the reduced data and sends it to the memory control circuit 123. Thereby, the memory control circuit 123 stores the binarized data in the reduced image memory 124. The rectangular division processing circuit 131 executes the division processing at the end of the storage processing described above.

The rectangular division processing circuit 131 converts the rectangular data obtained as a result of the processing into a buffer memory control circuit 133.
, Is temporarily stored in the rectangular data buffer memory 134, and is read and written during the integration processing. When the rectangle division processing circuit 131 completes the creation of the list of rectangle data after the integration processing, the extraction processing circuit 132 reads out multi-valued pixel data via the page memory control circuit 141.

The multi-valued pixel data has been written to the page memory 142 by the main scan by the time the reference of the extraction processing circuit 132 is started.

The extraction processing circuit 132 is provided in the page memory 14
2, the multi-valued pixel data is read out by random access, the multi-valued pixel data in the rectangular area obtained by the rectangular data is sent to the image discriminating circuit 135, and the image data outside the rectangular area is converted to the high-efficiency coding circuit for the background. 154. The image determination circuit 135 determines the image characteristics in the rectangular area,
The determination result is added to the rectangular data via the buffer memory control circuit 133 and stored in the rectangular data buffer memory 134. The image discriminating circuit 135 sends the signal to the low-frequency image high-efficiency coding circuit 152 or the high-frequency image high-efficiency coding circuit 153 based on the discrimination result.

The encoded data compressed by the high-efficiency encoding circuit for low-frequency image 152, the high-efficiency encoding circuit for high-frequency image 153, and the high-efficiency encoding circuit for background 154 are transmitted through the encoded data storage control circuit 155. It is stored in the main body memory 160.

Upon completion of the high-efficiency encoding process, the rectangular data compression circuit 156 compresses the rectangular data and stores it in the main body memory 160 under the control of a main body control circuit (not shown). Thus, the storage area of the main body memory 160 is managed by the main body control circuit, and data collision is controlled by controlling the data storage area and performing an overflow check.

When an image output command is issued, the encoded rectangular data is decoded and the rectangular data buffer memory 134 is decoded.
To save temporarily. The image expansion control circuit 171 reads the encoded data from the main memory 160 with reference to the buffer memory for rectangular data, and decodes the low-frequency image
High-frequency image decoding circuit 173, background decoding circuit 17
4 to perform appropriate decoding processing for each of characters, photographic images, and background. Such image data is stored in the page memory 142 via the developed image storage control circuit 175 and transmitted from the output interface 180.

Next, the processing operation of the rectangular division processing circuit 131 in this embodiment will be described in detail.

The processing operation of the rectangular division processing circuit 131 in the present embodiment will be described with reference to FIGS.

FIG. 2 is a flowchart showing a schematic process in the rectangular division processing circuit 131. The rectangular division processing circuit 131 reads out the binary data as the reduced image from the reduced image memory 124, determines whether the binary data belongs to the image area (step 1), If it is determined that the data belongs to the image area, it is confirmed whether a pixel other than the previously extracted rectangular area has been found (step 2).

If the rectangular division processing circuit 131 determines in step 2 that the pixel is outside the rectangular area, it stores the coordinates of the pixel as a black pixel start position (step 3).

The rectangle division processing circuit 131 determines the search direction at the black pixel start position, and dynamically changes the search range from the maximum search distance in the determined direction to perform the search (step 4).

When detecting the diagonal point position, the rectangular division processing circuit 131 replaces the X coordinate of the black pixel start position with the diagonal point position in the case of a search to the left (step 5).

The rectangular division processing circuit 131 determines whether or not the rectangular area determined by the black pixel start position and the diagonal point position overlaps another rectangular area already obtained (step 6).

If it is determined in step 6 that the rectangular areas overlap, the rectangular division processing circuit 131 obtains rectangular data of a rectangular area obtained by integrating the overlapping rectangular areas (step 7). Here, the integration processing means that the rectangular data is stored in the rectangular data buffer memory 134 as a rectangular list having pointers in both directions. Therefore, when new rectangle data is added to the rectangle list, if the overlap of the rectangular areas is detected by comparing it with the other rectangle data, it is integrated into the maximum circumscribed rectangle of the overlapped rectangles. This integration process is performed recursively until the overlapping rectangular data is eliminated.

On the other hand, if the rectangular division processing circuit 131 determines in step 6 that they do not overlap, the rectangular area data is added to the rectangular list in the rectangular data buffer memory 134 by the buffer memory control circuit 133 (step 8).

When the rectangular division processing circuit 131 returns to step 1 and determines that the image area is not raster-scanned, it determines that a list of rectangular area data has been obtained (step 1) and ends this routine.

Steps 1 to 8 described above are the schematic operations of the rectangular division processing circuit 131.

Next, the algorithm for determining the search direction and the dynamically changed search range in step 4 shown in FIG. 2 will be described with reference to FIGS.

In the search operation of the rectangular area in the present embodiment, the specified distance is changed during the search operation using a searcher having a search range corresponding to the specified distance in the horizontal and vertical directions centering on the point of interest. In the meantime, the search is performed in the search range around a point separated by a specified distance in the horizontal and vertical directions from the point of interest of the searcher.

Next, the operation of determining the search range of the search element in step 4 shown in FIG. 2 and the subsequent operation of moving the search element will be described with reference to FIGS.

FIG. 3 is a schematic diagram showing an algorithm for determining the search direction of the searcher in step 4 of FIG.

In FIG. 3, * is the pixel of interest, and the rectangular division processing circuit 131 has the reference numerals 5, 6, 7, and 8 among the eight pixels near the pixel of interest * as indicated by the bold line in the figure. The image is divided into a left area composed of adjacent pixels and a right area composed of adjacent pixels denoted by reference numerals 1, 2, 3, and 4, and a total of right-total and left-tot of pixel values of each area.
Find al. The rectangular division processing circuit 131
-Total and Left-Total are compared, and the direction of the region having the larger total value is determined as the search direction.

FIG. 4 is a schematic diagram showing a search range of a search element.

In FIG. 4, * indicates the pixel of interest shown in FIG. The numbers 1, 2, 3, and 4 in the grid indicate the search distance r, and the pixels with such numerical values can be the search center point (cx, cy) at each search distance.

If the operation of determining the search range in step 4 in FIG. 2 is generalized, it becomes (x + r + 1, y ± t).
If r is an even number, t = 2 * t-1 (k is a natural number obtained by r / 2), and if r is an odd number, t = 2 * t-1.
* K (k is an integer including 0 obtained by [r / 2]. Here, if r / 2 = i, then [i] indicates the largest integer not exceeding i. If r is 3, then k
= 1. ).

Therefore, according to the above formula, if the search distance r is 1, for example, eight neighboring pixels that are two pixels away from the target pixel in the horizontal or vertical direction are determined as the search range. Assuming that the search distance r is 2, eight search centers each of two search center points separated by one pixel in the vertical direction from three pixels separated in the horizontal direction are determined as the search range. Eight pixels near the search center point thus obtained, which are indicated by arrows, are the search range.

FIG. 5 is a schematic diagram showing an example of the processing operation shown in FIG.

FIG. 5A is a schematic diagram showing the state of the rectangular division processing circuit 131 in a state where the black pixel start position is stored. In this state, the upper left vertex of the rectangular area is detected. The direction has not been determined yet, and the case where the search distance r is disclosed from 2 is shown. In such a case, if the algorithm shown in FIG. 3 is executed and the sum of the pixel values of the search range r = 2 is set to 1 for the black pixel, T
_H (this is an abbreviation for totalH) = 5,
T _V (which is an abbreviation of totalV.) = 8 and since, because it is T _H <T _V and T _H ≠ 0, and the flow proceeds in the horizontal direction by a search distance r.

FIG. 5B shows a state in which the search direction is determined to the right by the algorithm shown in FIG.
= 2 has been moved.

FIG. 5C shows a state in which the process of determining the search direction and moving the searcher has been repeated twice from the state of FIG. 5B. Therefore, FIG.
In each of (c), the horizontal movement is performed to the right by the search distance 2. In the state shown in FIG. 5C, T _H = 0 and T _V = 15
Since T _V > 0 and T _H = 0, the search distance r = 2
It will proceed in the vertical direction as it is.

FIG. 5 (d) shows a state in which the object is vertically moved downward at the search distance r = 2 from the state of FIG. 5 (c).

FIG. 5 (e) shows a state in which the object is vertically moved rightward by a search distance 2 from the state of FIG. 5 (d).

FIG. 5 (f) shows a state where the object has been moved vertically downward twice at the search distance 2 from the state of FIG. 5 (e). In this state, since T _H and T _V = 0, the search distance r is changed by subtracting 1 from the search distance r.

FIG. 5 (g) shows a state in which the search distance r is changed to 1 since there is no black pixel in the search range in the vertical and horizontal directions in FIG. 5 (f). This is the process of dynamically changing the search range in the present embodiment. In this manner, a diagonal point is detected in the present embodiment.

Next, an algorithm of software for realizing the processing described with reference to FIG. 5 will be described with reference to FIGS.

FIG. 6 is a flowchart showing a rectangular area search operation in the rectangular division processing circuit 131.

The rectangular division processing circuit 131 divides the binary image stored in the reduced image memory 124 into rectangular areas by executing the binary image. When it finds a black pixel outside the other rectangle, it starts changing the dynamic search range.

The rectangular division processing circuit 131 checks whether the search range of the search element is larger than 0 (step 101).
If it is determined in step 101 that the search range is larger than 0, it is checked whether the current position (sx, sy) of the search element is within the image area (step 102).

If it is determined in step 102 that the search element is not within the image area, the rectangular division processing circuit 131 checks whether the current position (sx, sy) is different from the black pixel start position (step 119). If the determination in step 119 is affirmative, the current position (sx, sy), which is the diagonal point of the black pixel start position, is stored as the diagonal point position as a search result (step 120). Thereby, the rectangular division processing circuit 1
In step 31, since the rectangular data has been obtained, this routine ends.

On the other hand, if it is determined in step 102 that the search element is within the image area, totalH, total
Calculate IV (step 103).

The rectangular division processing circuit 131 determines in step 10
In steps 4 to 108, the values of totalH and totalV are determined to determine whether to change the search direction. Specifically, in step 104, totalH and totalV
Are equal to each other, and in step 105, totalH
It is checked whether <totalV and total | H | = 0, and in step 106, totalV <total
H and total | V | = 0, and in step 107, totalH is 0 or less and total
It is checked whether l | V | = 0.
Check whether talV is greater than 0 and totalH is 0.

The rectangular division processing circuit 131 determines in step 10
4. If a positive determination is made in step 105 or step 107, the search direction of the searcher is changed from the vertical direction to the horizontal direction (steps 110, 111, and 113), and the search direction is positive in step 106 or step 108. If it is determined, the search direction of the searcher is changed from the horizontal direction to the vertical direction (steps 112 and 114). If the determination is negative in step 108, the search direction is left unchanged (step 109). .

The rectangular division processing circuit 131 determines in step 11
It is confirmed whether the search direction determined in step 5 is the horizontal direction (step 115).

If the determination in step 115 is affirmative,
A horizontal search process is performed (step 116), the search range is changed (step 118), and the process returns to step 101. If the search range is smaller than 0 in the determination of step 101, the coordinates (sx, sy) of the target point at that time are stored, the search process ends, and the present routine ends.

If the determination in step 115 is negative,
A vertical search process is executed (step 117), the search range is changed (step 118), and the process returns to step 101. If it is determined in step 101 that the search range is smaller than 0, this routine ends.

FIG. 7 is a flowchart showing a horizontal search operation subroutine.

This flowchart corresponds to step 116 in FIG.
3 shows details of the processing contents.

When this subroutine is started, the rectangular division processing circuit 131 checks whether the search direction is left (step 201), and sets a horizontal search range Xrange (steps 202 and 203).

Specifically, the rectangular division processing circuit 131
If it is determined in step 201 that it is the left direction, a value obtained by multiplying the current search range by “−1” is substituted into the horizontal search range Xrange (step 202), and if it is determined in the step 201 that it is the right direction. , Horizontal search range Xrange
Is left as it is (step 203).

If the search range (sx + Xrange, sy) of the search element is within the image area (step 204), the rectangular division processing circuit 131 calculates totalH and totalV (step 205).

The rectangular division processing circuit 131 determines in step 20
In steps 6 to 208, the next search direction is determined based on the values of totalH and totalV (steps 209 to 211), and the search element is moved (step 21).
2).

More specifically, in step 206, total
It is checked whether H and totalV are not both 0, whether totalH is equal to totalV in step 207, and whether totalH <totalV is checked in step 208. In step 206, totalH, t
If it is determined that both the values of totalV are 0, the routine is terminated and the process returns to step 116 shown in FIG.

If a positive determination is made in step 207, the search direction is left as it is (step 209), and s
By adding Xrange to x (step 212), the search element is moved, the present routine ends, and the process returns to step 116 shown in FIG.

If an affirmative decision is made in step 208, the search direction is changed to the horizontal direction (step 21).
0), add only Xrange to sx (step 21)
2) Then, the search element is moved, the present routine ends, and the process returns to step 116 shown in FIG.

FIG. 8 is a flowchart showing a subroutine for a vertical search operation.

This flowchart corresponds to step 117 in FIG.
3 shows details of the processing contents.

When this subroutine is started, the rectangular division processing circuit 131 substitutes the current search range into the vertical search range Yrange (step 301).

If the search range (sx, sy + Yrange) of the search element is within the image area (step 302), the rectangular division processing circuit 131 calculates totalH and totalV (step 303).

The rectangular division processing circuit 131 determines in step 30
In steps 4 to 306, the next search direction is determined based on the values of totalH and totalV (steps 307 to 309), and the search element is moved (step 31).
0).

Specifically, in step 304, total
It is checked whether both H and totalV are not 0, it is checked in step 305 whether totalH and totalV are equal, and in step 306, it is checked whether totalH <totalV. In step 304, totalH, t
If it is determined that both the values of totalV are 0, the present routine is terminated and the process returns to step 117 shown in FIG.

If a positive determination is made in step 305, the search direction is left as it is (step 307), and s
By adding Y only to y (step 310), the search element is moved, the present routine is terminated, and the process returns to step 117 shown in FIG.

If a positive determination is made in step 306, the search direction is changed to the horizontal direction (step 30).
8) Add only Yrange to sy (step 31)
0), the search element is moved, the present routine ends, and the process returns to step 117 shown in FIG.

If a negative determination is made in step 306, the search direction is changed to the vertical direction (step 30).
9) Add only Yrange to sy (step 31)
0), the search element is moved, the present routine ends, and the process returns to step 117 shown in FIG.

As described above, according to the data processing circuit of the present embodiment, not only the character area and the photographic image can be extracted from the original image, but also the background area can be extracted from the original image. Since image data can be handled as continuous pixel data of a constant luminance level, compression efficiency of the image data can be improved, and a compression method suitable for a rectangular area can be selected, so that image quality is not reduced. The compression ratio can be improved. The searcher moves over a small gap such as between characters or character strings. This also enables high-speed processing.

According to the data processing circuit of the above-described embodiment, in addition to the above-described effects, a rectangular area can be specified by obtaining one vertex and its diagonal vertex. The search is performed only in the horizontal and vertical directions, and at the beginning of the search, the search range of the searcher is increased, and the search range of the searcher is reduced near the apex of the rectangular area. It can be changed dynamically. Therefore, the search processing of the present invention can accurately and quickly search near the vertices of the rectangular area.

[0112]

According to the first or third aspect of the present invention, not only the character region and the photographic image are extracted from the original image but also the background region is extracted from the original image. , The background image data can be treated as continuous pixel data of a constant luminance level, so that the compression efficiency of the image data can be improved and a compression method suitable for a rectangular area is selected. Therefore, the compression rate can be improved without deteriorating the image quality. The searcher moves over a small gap such as between characters or character strings. This also enables high-speed processing.

According to the second aspect of the present invention, in addition to the above-described effects, by providing the above configuration, a rectangular area can be specified by obtaining one vertex and its diagonal vertex. The search is performed only in the horizontal and vertical directions, and at the beginning of the search, the search range of the searcher is increased, and the search range of the searcher is reduced near the apex of the rectangular area. It can be changed dynamically. Therefore, the search processing of the present invention can accurately and quickly search near the vertices of the rectangular area.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a data processing circuit according to the present embodiment.

FIG. 2 is a flowchart showing a schematic process in a rectangular division processing circuit 131;

FIG. 3 is a schematic diagram showing an algorithm for determining a search direction of a searcher in step 4 of FIG. 2;

FIG. 4 is a schematic diagram showing a search range of a search element.

FIG. 5 is a schematic diagram showing an example of the processing operation shown in FIG. 2;

FIG. 6 is a flowchart showing a rectangular area search operation in the rectangular division processing circuit 131;

FIG. 7 is a flowchart showing a horizontal search operation subroutine.

FIG. 8 is a flowchart showing a vertical search operation subroutine.

[Explanation of symbols]

 111 Scanner I / F 112 Scan Buffer 121 Reduction Processing Circuit 122 Binarization Processing Circuit 123 Memory Control Circuit 124 Reduced Image Memory 131 Rectangular Division Processing Circuit 132 Extraction Processing Circuit 133 Buffer Memory Control Circuit 134 Rectangular Data Buffer Memory 135 Image Discrimination Circuit 141 Page memory control circuit 142 Page memory 152 High-efficiency encoding circuit for low-frequency image 153 High-efficiency encoding circuit for high-frequency image 154 High-efficiency encoding circuit for background 155 Encoded data storage control circuit 156 Rectangular data compression circuit 160 Main unit Memory 171 Image expansion control circuit 172 Low frequency image decoding circuit 173 High frequency image decoding circuit 174 Background decoding circuit 175 Expanded image storage control circuit

──────────────────────────────────────────────────続 き Continued on front page (51) Int.Cl. ⁶ Identification code FI H03M 7/30 H04N 1/21 H04N 1/21 1/411 1/411 B41J 3/00 D

Claims (5)

[Claims] 1. A scanner for reading as multi-valued pixel data from an original image in which characters and photographic images are mixed, and reduced data indicating an average level of two-dimensionally adjacent multi-valued pixel data transmitted from the scanner. Reduction processing means for converting, binarization means for binarizing reduced data obtained from the reduction processing means, and a rectangular area on a page which is determined by determining a rectangular area on a page from the binary data obtained from the binarization means Rectangular data generating means for generating rectangular data of an area, a rectangular data memory for storing rectangular data from the rectangular data generating means, and a page memory for storing multi-valued pixel data from the scanner in page units; Image discriminating means for detecting image characteristics from multi-valued pixel data read from the page memory included in a rectangular area defined by the rectangular data; The multi-valued pixel data read from the page memory based on the determination result of the image determining unit is a low-frequency image high-efficiency encoding unit, a high-frequency image high-efficiency encoding unit, and a background high-efficiency encoding unit. And the high-efficiency encoding means for the low-frequency image, the high-efficiency encoding means for the high-frequency image, and the high-efficiency encoding means for the background. A data processing apparatus comprising: storage control means for storing; and a plurality of expansion processing means for expanding high-efficiency code data read from the main body memory into image data. 2. The data according to claim 1, wherein said rectangular data generating means uses a search element for dynamically changing a search range, and performs integration processing so as not to cause overlapping of the divided rectangular areas. Processing equipment. 3. The data processing apparatus according to claim 1, wherein said image discriminating means detects an image characteristic from a difference value between adjacent pixel data. 4. The high-efficiency encoding means for high-frequency image is JPEG compression suitable for an image having a small change between adjacent pixels and having high correlation between pixels. The compression is suitable for an edge image in which there are many changes between adjacent pixels and the correlation between the pixels is low, and the background high-efficiency encoding means encodes continuous constant-level pixel data with high efficiency. The data processing device according to claim 1, wherein: 5. The data processing apparatus according to claim 1, wherein said rectangular data is encoded with high efficiency and stored in a main body memory.