JPH05122535A - Image processor - Google Patents

Abstract

PURPOSE:To provide the image processor improving compression efficiency while keeping satisfactory picture quality. CONSTITUTION:The base of a source image signal is detected by a base detection part 1, and a base level and a value subtracting the base level from the source image signal are respectively separately encoded by a multilevel image compression part 3 and a binary image encoding part 5. At such a time, by judging the color of the base, an encoding method is changed.

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 image compression, and more particularly to the technical field of highly efficient compression and transmission of halftone images.

[0002]

2. Description of the Related Art Conventionally, MH has been used as an image compression technique.
There are binary black and white image compressions such as coded MR and MMR coding, which are used for facsimiles and the like. Further, it has been considered that this binary white / black compression should be applied to each color component for a color image. However, the color image is
Due to its nature, if sufficient gradation cannot be obtained in a binary image, its significance is lost. In addition, even for black and white images, the tendency to emphasize the gradation reproducibility has recently become stronger. Therefore, the systematic dither method, the error diffusion method, or the like is used as the binarizing means, and the gradation is produced by the pseudo gradation expression.

[0003]

However, the binary image represented by the pseudo gradation has a very low compression efficiency, and in the binarization method using the error diffusion method, the compression rate exceeds 1.0 and the original data is It may even increase the data capacity. Even if the image is not an image containing halftones, if there is a fog in the background part, such as a newspaper manuscript,
It has a great influence on the compression rate. To prevent this, one of the solutions is to remove the density component of the background portion. However, in color documents, the background color and density component have a very important meaning. What you do cannot be a solution.

Further, even if a document mainly composed of characters is compressed by using a multivalued image compression method based on ADCT (Discrete Cosine Transform), the compression efficiency is not improved at all, and the peripheral portion of the characters is not compressed. Problems such as deterioration of image quality also occur.

On the other hand, recently, from the standpoint of environmental protection, the use of recycled paper has become more frequent, but in the case of image transmission, it may be unnecessary to transmit the color of the background portion.

Further, the importance of the background color may be determined depending on whether the input image is color or black and white.

Therefore, it is an object of the present invention to provide an image processing apparatus using an encoding method with which image data compression efficiency is good.

Another object of the present invention is to provide an image processing apparatus having good operability.

[0009]

In order to solve the above-mentioned problems, the image processing apparatus of the present invention comprises a judgment means for judging whether the input image is a color image or a monochrome image,
It is characterized by further comprising: detection means for detecting a background portion of the input image; and encoding means for encoding the input image based on the determination result by the determination means and the detection result by the detection means.

Further, a detecting means for detecting a background portion of the input image, a removing means for removing the color of the background portion from the input image, a first mode for performing the removal by the removing means, and a second mode not performed. And a coding means for coding the input image according to the mode selected by the selecting means.

[0011]

EXAMPLES In the examples of the present invention described below, the background density portion is separated from the data component of the image, and two types of images, that is, the background portion image and the image without the background are generated, The image compression efficiency of the part excluding is very high.

Further, as for the background portion, data is stored or transmitted in another compression means or in a non-compressed form, so that the data can be reproduced without losing the background portion when the data is decompressed. It becomes effective.

FIG. 1 shows a first embodiment of the present invention. An original image signal is first input to a background detecting section 1 to obtain a background level or compression element data closely related to the background level. , The subtractor 2 and the multi-valued image compression unit 3. The compressed element data is compressed by the multi-valued image compression unit 3 and output to the synthesis unit 6. The background level detected by the background detection unit 1 is subtracted from the original image signal by the subtractor 2, the background component is removed from the original image signal, and the image with the background component removed is binarized by the binarization unit 4. Be converted. The output of the binarization unit 4 is encoded by the binary image encoding unit 5, combined with the multi-valued image compressed data by the combining unit 6, or simply added, and then output from the data transmission unit 7.

FIG. 2 shows an example of the background detecting section 1. The original image signal is first input to the DCT conversion unit 11, and then the DC component extraction unit 12 extracts the DC component of the image. The portion consisting of the DCT conversion unit 11 and the DC extraction unit 12 is referred to as an average value calculation unit 10, and the average value of the image is calculated. This output signal is output by the maximum value limiting unit 13 described later.
If it is equal to or higher than a certain fixed background level, it is replaced with the fixed value (background level).

After that, the output data of the maximum value limiting unit 13 is input to the latch unit 14 and the difference calculating unit 15, and in the difference calculating unit 15, the difference from the previous pixel latched in the latch unit 14 is the compressed element data. Is output as. The output of the maximum value limiting unit 13 is output as the background level.

FIG. 3 shows another example of the background detecting section 1. The original image signal is smoothed in the smoothing unit 16 by, for example, a smoothing filter having coefficients as shown in the figure. By having this smoothing filter, the compression efficiency is improved without being affected by noise or the like. Next, the maximum value limiting unit 13 limits the maximum value of the background level, which becomes the compression element data and the background level.

FIG. 4 shows another example of the background detecting section 1. The original image signal is first input to the latch 18 and also to the comparator 19. Latch S at the beginning of one image
The maximum value is set by the ET signal. Latch 18
The output of is compared in size by a comparator 19, and if the original image signal is smaller, a trigger signal is generated for the latch 18 and the original image signal is latched in the latch 18. In this way, when the entire one surface of the image flows, the minimum value is latched in the latch 18. This minimum value is obtained as the compression element data and the background level by limiting the maximum value in the maximum value limiting unit 13.

The data input to the maximum value limiting unit 13 is
It is input to the comparator 20 and the selector 21. On the other hand, the comparator 20
If the maximum background level is set and the input data exceeds the maximum background level, the selector 21 is controlled to output the maximum background level. If the input data is less than the maximum background level, The selector 21 controls to select and output the input data.

The background level can be detected by using another method such as taking a histogram and taking the maximum frequency value.

FIG. 5 shows an example of a circuit configuration for decompressing and decoding code data on the receiving side when the above-described image compression coding is performed. The data received by the data receiving unit 31 is separated by the data separating unit 32 into binary encoded data and multilevel compressed data.

The binary coded data is converted into binary data by the binary image decoding unit.
Value data, and the multi-value compressed data is converted to multi-value image decompression unit 34.
Is expanded to multi-valued data. These two types of data are combined into one data by the data combining unit 35 and become final data or final output.

FIG. 6A is a block diagram for implementing the data synthesizing unit 35.
This is an example, and the binary data is converted into a multivalued image by the multivalued image restoration unit 41. For example, a smoothing filter is used for the inside, or simply 1/0 of binary is 2 if it is 8-bit multivalued.
55/0 is also an example. The multi-valued restored data and the multi-valued data are added by the adder 42 to obtain multi-valued data. This multi-valued data is re-binarized by the binarization unit 43, and the selection unit 44 selects either a binary image or a multi-valued image as desired (depending on whether the printer is a binary printer or a multi-valued printer). But you can use the one you need.

FIG. 6B is another embodiment for realizing the data synthesizing unit 35, and multivalued data is binarized by the binarizing unit 45. The binary data and the output of 45 are input to gates 46 and 47. At gate 46, the two images are ORed together. If the two images are both "1" at the gate 47, the two images overlap, and the output of 47 for detecting the overlap and counting up the counter 48 controls the count up / down. If the output of 47 is "1", it instructs the gate 51 to generate a count enable signal. The comparator 49 determines whether the number of counters matches “0”,
When the output of the gate 46 is "0" and the counter 48 is other than "0", the counter counts down "1" and the output "0" of the gate 46 is set to "1" by the gate 50. In this way, the number of times "1" of two images overlaps is counted by the counter 4
8 counts, and when both images are both “0”, the operation to generate “1” and reduce the number of overlaps is prevented, and the number of dots is prevented from decreasing when two images are overlapped. It becomes possible to save.

FIG. 6C shows a simple implementation example for implementing the data synthesizing unit 35, which is a form obtained by cutting out a part of FIG. 6B.

The multi-valued data is binarized by the binarization unit 61 and then ORed with the binary data by the gate 62.
In this case, the density is not stored, but if the binarization method is originally a density-non-conservation type such as the systematic dither method, the density information has already been lost. However, the effect is small, and the configuration shown in FIG. 6C is sufficient.

FIG. 9 shows the multi-valued image decompression unit 3 of FIG. 5 when the compression element data shown in FIG. 2 is compressed by the multi-valued image compression unit 3.
4 is an example of the configuration. First, the input is the difference expansion unit 91,
The output, which is data-decompressed, is a difference, so the addition unit 92
Is added with the data of the previous pixel. The previous pixel is latched by the latch 93, but the output of the adder 92 is latched by the latch 93 for processing by the next pixel again. The latch 93 of the first pixel of one screen is cleared, and "0" is added to the output of the adder unit 92 to be output.

FIG. 7 shows another embodiment for realizing the background detecting section 1. The original image signal is given as an address of the frequency table 73 via the selector 72. Frequency table 7
3, "0" is written as an initial value in all addresses, but the data of the designated address is read out, the adder 75 adds "1", and the result is written in the frequency table 73 via the buffer Buf74. For one pixel, the pixel clock CLK reads the frequency table 73 in the first half and disables the Buf 74. In the latter half of CLK, the frequency table 73 is written to enable the Buf 74.

When such an operation is repeated for one screen, FIG.
As shown in A, the appearance frequency distribution for each original image signal level is obtained. FIG. 8A shows two frequency distribution examples with a broken line and a solid line.

After the above operation is completed and the frequency distribution is obtained, the counter 71 generates data instead of the original image signal, and the data is given to the address of the frequency table via the selector 72. At this time, the frequency table is read-only and is given to the adder 76. Adder 7
The output of 6 is latched in the latch 77 for each pixel (for each clock of the counter), and the output is added to the adder 76 again. Therefore, the adder 76 latch 77 can obtain the cumulative result of addition.

The output of the latch 77 is compared with a threshold th in a comparator 78, and when the value exceeds one value or exceeds th, a latch clock is generated for the data latch 79,
The output value of the counter 71 at that time is latched in the data latch 79. This value becomes a background level candidate signal, but the maximum value limiting unit 80 limits the maximum value and outputs it as the background level or compressed element data. The relationship between th and the background level candidate latched by the data latch 79 in this case is shown in FIG. 8B. The sum of all frequencies from the original image level 0 to the candidate value of the background level corresponds to the area shown by the diagonal lines in the figure,
The area becomes the threshold th.

Now, one embodiment of the synthesizing unit 6 shown in FIG. 1 will be briefly described. Here, it is assumed that the binary image coding unit uses MH coding, which is one-dimensional coding, and the background detection unit 1 performs detection as shown in FIG.

Therefore, the multilevel image compression unit for the background level replaces the background level UL with the run length for encoding. The codes are, for example, white run length UL, black run length 0, white run length (LL-UL) EOL (end
It may be set as "of line". Where L
L is the run length of one line. Therefore, the code is all white when decoded, but if the decoder knows this agreement, the background level is the value of "UL", and if the background removal is not necessary, the background level is added. Decoding may be performed, and if background removal is necessary, it may be ignored. Of course, a decoder that does not know the underlying coding agreement can reproduce the image with the background removed. Therefore, in the synthesizing unit 6, the output of the multi-valued image compressing unit 3 may be added to the beginning of the output of the binary image encoding unit 5. A special combination of codes may be prepared to determine whether the first line at the beginning is a background level code or a normal binary image compression result. For example, white 0, black 0, white 1, black 0, white 2, black 0, white 3, black 0, white 4, black 0, ..., White n-1, black 0, white n, black 0, white UL ,
It may be black 0 and white (LL-UL-n (n + 1) / 2).

Further, in all of the present embodiments, it is possible to omit the multivalued image compression section 13 and send raw data because of the small amount of information of the background image portion.

As described above, according to the above-described embodiment of the present invention, the image is divided into the background portion (background density level) and the remaining portion of the image.
It is possible to obtain an image with a high compression rate without being affected by the background level even in the case of a character image that has been binarized by the error diffusion method, etc. is there.

FIG. 12 is a diagram showing another embodiment. The background level and compression element data output from the background detection unit 1 are input to the background processing unit 100, and the original image signal is also input to the background processing unit. It The background processing unit 100 determines whether the background image is color or black and white, and based on the determination result, the background level,
The compressed element data is processed and output.

FIG. 10 is a block diagram for realizing the background processing section 100. The input background level is subtracted from the original image signal by the subtractor 113, and is input to the background determination unit 101. This difference is compared by the background determination unit 101 with a threshold value th10, and th1
If it is smaller than 0, it is determined to be a base portion. As a result of this determination, the counter 102 counts the number of times the background portion is determined. On the other hand, the chromaticity calculation unit 103 calculates the chromaticity of the original image signal. In the calculation here, first, the image data of R, G, B is converted into L ^* , a ^* , b ^*.

[0037]

[Outer 1] The value of is output. Instead of this value | a ^* | + | b ^* |
May be output, Y, IQ, L, U, V,
A color space such as Y, Cr, Cb may be used. The output of the chromaticity calculator is compared with the threshold value th12 by the comparator 104. If it is larger than th12, it is determined to be a color pixel and is input to the clock input of the counter 105. The counter 105 counts the number of color pixels that are the background in order to allow the counter 105 to count only pixels that are determined to be the background by the output of the background determination unit 101.

When the image is pre-scanned as described above, the divider 106 obtains the result of dividing the counter 105 by the counter 102, that is, the ratio of determining the background image as color. This ratio is compared with the threshold value th11 by the comparator 107, and if it is th11 or more, it is determined that the background portion of the processed image is a color background. When the background is color, the input background level and the input compression element data are cleared by the clear circuit 11.
0, cleared to “0” by the clear circuit 109. Therefore, when the background color is color, the entire background including the background color is binarized and encoded, and multi-value compression is not performed.

If the background color determination data is canceled by the cancel circuit 108, the input background level and the input compression element data are output without any processing.

In this case, the background color determination data output from the background processing unit 100 is added to the combining unit 6 and transmitted. In that case, as shown in FIG. 13, the background color determination data is separated by the data separating unit 32 and input to the logic circuit 114. Even if the background removal signal is input in advance to the logic circuit 114 and the background is set to be removed, if the background color determination data indicates a color, the background is not removed. The background is removed only when the background removal signal is set to remove the background and the background color determination data indicates that the background is not color. The background is removed by clearing the output data of the multivalued image decompression unit 34.

FIG. 11 shows another embodiment for realizing the background processing section 100. The output of the comparator 107 is the cancellation circuit 1
The operation in the case of being canceled in 08 is as described above.

If not canceled, the background color determination data is input to the selectors 111 and 112 and becomes a control signal for the selector. The input background level and the original image signal are input to the selector 111, and the input compression element data and the original image signal are input to the selector 112.
When the background color determination data indicates black and white, the background level and compression element data are selected by the selectors 111 and 112 and output as they are, and the subsequent operation is as already described. When the background color determination data indicates color, both the selector 111 and the selector 112 select the original image signal and output it as output background level output compression element data. Therefore, in the subtractor 2 of FIG. 12, the original images are subtracted from each other and the output is always 0. Further, the multi-valued image compression unit 3 compresses the original image data as it is. Therefore, the multi-valued image compression unit 3 may use ADCT compression or the like.

As described above, according to the above embodiment of the present invention, the image is divided into the background portion (background density level) and the remaining portion of the image.
It is possible to obtain an image with a high compression rate without being affected by the background level even in the case of a character image that has been binarized by the error diffusion method by separating the images of one element and encoding and compressing them separately. is there. Further, it is possible to save the background when the background image is in color and to remove it when the background is in black and white.

FIG. 14 is a block diagram showing another embodiment of the present invention.

In FIG. 14, reference numeral 301 denotes an image input section composed of a CCD sensor, 302 denotes a background processing section for performing the processing described later, and 303 denotes color component signals R ″ and G ″ whose background data has been removed. , B ″ is converted to a luminance signal L ^* and chromaticity signals a ^* and b ^* , and 304 is a converted L
ADCT of JPEG for each of ^* , a ^* , and b ^* components
(Adaptive Discrete Cosine
ADCCT compression unit for performing image data compression by the Transform method, 305 is a background level compression unit for compressing (coding) the background level data R ′, G ′, B ′ extracted by the background processing unit 302, and 306 and 307 are ADCT compression section 304 and background level data compression section 30 respectively
5, a memory that stores the output data from the memory 5, 308, a synthesizing unit that synthesizes the data stored in the memories 306 and 307,
309 is a data transmission unit for transmitting the combined data, 3
Reference numeral 10 is a system controller for controlling the above-mentioned respective units, and 311 is an operation unit having a key input means for performing mode setting and the like.

The details of the base processing section 302 are shown in FIG.

The structure of the background color determination unit 321 is shown in FIG.
The description is omitted because it is the same as that shown in FIG. Figure 1
5, in addition to the configuration of FIG. 10, 1 based on the monochrome / color discrimination result for each pixel output from the comparator 104.
A document color determination unit 322 is provided that determines whether the input image for the screen is a monochrome image or a color image. In the subtractor 323, the background level data is subtracted only from the pixels determined to be the background for the input image data R, G, B, and the background removed image data R ″, G ″, B ′. 'Is output.

FIG. 16 shows a control procedure by the system controller 310 in the above configuration.

First, in step S100, it is determined whether or not the original color determination mode (ACS mode) is selected by the operation unit 311. If the ACS mode is selected, the input image is a monochrome image or a color image. It is determined (S101). On the other hand, if the ACS mode is not selected, the process proceeds to step S103. ACS
When the mode is selected, the document color determination unit 3
The number of pixels determined to be chromatic by a counter provided in 22 is counted. If the count value is less than or equal to a predetermined value, it is determined to be a black-and-white document, and in the opposite case, it is determined to be a color document. This determination result is sent to the system controller 311 as document color determination data.

If the ACS determines that the document is in color, then in step S103, it is determined whether the background color is color. When the background color is color, control is performed so that the color background color data (R ', B', G ') is transmitted (S105). That is, the system controller 310 outputs an instruction signal to the background level compression unit 305 to output the background level data.

On the other hand, if the background color is black and white, then it is determined whether the background removal mode is selected by the operation unit 311 (S104). If the background removal mode is not selected, the background removal mode is selected. The density data y is controlled to be transmitted (S105). When the background removal mode is selected, the background color is controlled not to be transmitted (S10).
6). That is, the system controller outputs an instruction signal to the background level compression unit 305 so as not to output the background level data.

If it is determined by ACS that the original is black and white, it is determined whether the background removal mode is selected (S107). If it is selected, the background is removed (S106). If it is not selected, a black and white background is transmitted (S108).

The above ACS and base color determination are performed in the prescan operation by the image input unit (for example, CCD scanner) 301.

When the image input unit 301 includes a memory capable of storing image data for one image, for example, the memory is read for pre-processing separately from the image processing for transmission from the memory. The ACS and background color determination is performed.

FIG. 17 shows a modification of the embodiment shown in FIG.
In this embodiment, when the document is determined to be in color as a result of ACS, it is determined whether the background removal mode is selected, and if it is selected, the background is removed (S
106). On the other hand, if the background removal mode is not selected, the background color is determined (S202), if the background color is color, the color background is transmitted (S105), and if the background color is black and white, The base is removed (S106).

According to this embodiment, even if the background removal mode is not selected, the background transmission can be controlled according to the background color.

FIG. 18 shows a modification of the embodiment shown in FIG.
In this embodiment, when the document is determined to be in color as a result of ACS, the background transmission is performed (S105) depending on whether the background color is the background removal mode or not, regardless of whether the background color is color or not. Replace background removal (S106) (S
301).

When the original is color, the default is the background transmission mode, and when the original is black and white, the default is the background removal mode.

FIG. 19 is a block diagram showing the receiving side of the embodiment of FIG. In FIG. 19, 331 is a data receiving unit, 332 is multi-valued ADCT compressed data,
A data separation unit 333 for separating the background level data,
334 is a memory for storing the separated data, 33
5 decompresses the compressed data by ADCT, R ″,
An ADCT decompressor for outputting G ″, B ″ data, 33
Reference numeral 6 denotes a background level expansion unit 337 for extracting background level data R ′, G ′, B ′ from the compressed background level data.
Is a data combiner 338 for restoring the R, G, B values for each pixel from the multivalued data R ″, G ″, B ″ and the background levels R ′, G ′, B ′, and 338 is a laser beam printer. An image output unit including an inkjet printer, 3
39 is a system controller for controlling the receiving unit, 340
Is an operation unit for setting the background removal mode.

On the receiving side, the image data is restored and the image is output in the reverse procedure of the transmitting side.

The system controller determines the presence or absence of the background level data according to the data stored in the memory 334. If the background removal mode is designated by the operation unit 340, the background level data will be detected even if the background level data exists. The background level expansion unit 336 is controlled so that it is not output.

Instead of the ADCT compression unit 304 described above, another multi-value image data compression method may be used.

Further, the image output unit 338 may be a so-called bubble jet printer of the type described in US Pat. No. 4,723,129 which ejects droplets by utilizing film boiling due to thermal energy.

Further, in FIG. 15, the background color determination unit 3
In 21, it is also possible to determine whether the part other than the background is black and white or color. The background may not be removed when the portion other than the background is color, and the background may be removed when the portion is black and white.

That is, with the configuration as shown in FIG. 20, it is possible to determine the color / black and white of a portion other than the background and remove the background color according to the result.

Further, the system controller on the transmitting side is
When the receiving side can only output a monochrome image according to the protocol with the receiving side, the background color may be removed before transmission.

[0067]

As described above, according to the present invention, it is possible to obtain an image processing apparatus having an improved compression efficiency while maintaining a good image. In addition, it is possible to obtain an image processing device with good operability.

[Brief description of drawings]

FIG. 1 is a block diagram of an image encoding unit according to the present invention.

FIG. 2 is a block diagram showing a background detection unit.

FIG. 3 is a block diagram showing a background detection unit.

FIG. 4 is a block diagram showing a background detection unit.

FIG. 5 is a block diagram of an image decoding unit according to the present invention.

FIG. 6 is a block diagram showing a data synthesizing unit.

FIG. 7 is a diagram showing another example of a background detection unit.

FIG. 8 is a diagram showing a relationship between an original image signal level and an appearance frequency.

FIG. 9 is a diagram showing a multi-valued image decompression unit.

FIG. 10 is a diagram showing a background processing unit.

FIG. 11 is a diagram showing a background processing unit.

FIG. 12 is a block diagram of a modified embodiment of the image encoding unit of the present invention.

FIG. 13 is a block diagram of a modified embodiment of the image decompression unit.

FIG. 14 is a block diagram of another embodiment of the present invention.

FIG. 15 is a diagram showing a background processing section.

FIG. 16 is a diagram showing control by a system controller.

FIG. 17 is a diagram showing control by a system controller.

FIG. 18 is a diagram showing control by the system controller.

FIG. 19 is a block diagram of another embodiment of the present invention.

FIG. 20 is a block diagram of another embodiment of the present invention.

[Explanation of symbols]

 1 Background detection unit 2 Subtractor 3 Multi-value image compression unit 4 Binarization unit 5 Binary image encoding unit 6 Compositing unit 7 Data transmission unit 10 Average value calculation unit 11 DCT conversion unit 12 DC extraction unit 13 Maximum value limit Section 14 latch section 15 difference calculation section 16 smoothing section 18 latch 31 data receiving section 32 data separating section 33 binary image decoding section 34 multi-valued image expanding section 35 data synthesizing section 41 multi-valued image restoring section 73 frequency table 100 background Processing unit 101 Background determination unit 102 Counter 103 Chromaticity calculation unit 104 Comparator 105 Counter 106 Divider 107 Comparator 108 Cancel circuit 109 Clear circuit 110 Clear circuit 111 Selector 112 Selector 113 Subtractor 114 Logic circuit

Claims (4)

[Claims] 1. A judging means for judging whether an input image is a color image or a monochrome image, a detecting means for detecting a background portion of the input image, a judgment result by the judging means and a detection by the detecting means. An image processing apparatus comprising: an encoding unit that encodes the input image based on a result. 2. The image processing apparatus according to claim 1, wherein the encoding means encodes the input image using a plurality of encoding methods different from each other. 3. The image processing apparatus according to claim 1, wherein the plurality of encoding methods include multi-value data encoding and binary data encoding. 4. A detecting means for detecting a background portion of an input image, a removing means for removing the color of the background portion from the input image, a first mode for performing the removal by the removing means, and a second mode not performed. An image processing apparatus comprising: a selection unit that selects one of the modes, and an encoding unit that encodes the input image according to the mode selected by the selection unit.