JPH04139960A - Picture coder - Google Patents

Abstract

PURPOSE:To improve the picture quality and to enhance a coding efficiency by applying different coding to a line drawing part in plural colors extracted by an extraction means and other part than the line drawing part. CONSTITUTION:A signal output device 92 is configurated to be a shift register, receives a parallel 7-bit input signal 211 and outputs a serial signal in one bit each from the MSB. The output is a binary series signal D203. The signal output device 92 terminates a color signal of one picture element when the binary series signal is 1 or seven of Os are outputted and receives a succeeding input data. Moreover, the signal output device 92 outputs a signal Bt204 representing to which order bit of the binary series the bit outputted at present corresponds. Thus, when a 3-bit color character discrimination signal is subjected to binary series conversion and coded to 1-bit serial signal, the coding is implemented while preserving the color correlation and the coding efficiency is enhanced without separate coding of a 3-bit signal having correlation.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image encoding device that separates and encodes a line drawing part and a natural image part in a color image.

[Prior Art] Conventionally, as a method of encoding a color image, a method has been known in which the image is divided into blocks, orthogonal transformation is performed, and then the coefficients are quantized and encoded.

However, in the above-mentioned conventional example, since the coefficients after orthogonal transformation are quantized, high frequency components are lost, ringing occurs at the edges, and the quality of the text portion of the document is degraded.

Furthermore, when encoding at a low bit rate, block distortion occurs at the boundaries between blocks, making the outlines of characters unsightly.

On the other hand, in order to improve the quality of black characters, which are the most frequently used characters, the present applicant has proposed a technique in which black characters and other parts are separated and encoded.

However, since the above technique is configured to extract only one specific color of the original, there is room for improvement when, for example, there are characters in multiple colors in the original.

SUMMARY OF THE INVENTION Therefore, it is an object of the present invention to provide an image encoding device that can eliminate the drawbacks of the prior art, maintain good image quality, and has high encoding efficiency.

[Means and operations for solving the problem] In order to solve the above object, the image encoding device of the present invention includes an extraction means for extracting a plurality of color line drawing parts of input image data, and a plurality of lines extracted by the extraction means. The present invention is characterized by having an encoding means for performing different encoding on a color line drawing portion and a portion other than the line drawing portion.

In the above configuration, the extraction means extracts a multi-color line drawing part of the input image data, and the encoding means codes different codes for the multi-color line drawing part extracted by the extraction means and parts other than the line drawing part. make a change.

[Example code] Preferred embodiments of the present invention will be described below with reference to the drawings.

4. Impression cylinder FIG. 1 is a diagram showing the overall configuration of an image encoding apparatus according to a first embodiment of the present invention.

In FIG. 1, reference numeral 1 denotes an image input unit that inputs an image signal representing a document, which scans the document with a CCD line sensor, and scans the document by a CCD line sensor. It consists of an image sensor that outputs bit color component signals. Reference numeral 2 denotes an edge detection section, which detects high frequency component portions of the original image by a method described later.

3 is a color detection unit that detects pixels of a predetermined color component.

Reference numeral 4 denotes a color character determination section, which determines pixels of edge portions and predetermined color components. Reference numeral 5 denotes a binary series conversion unit, which converts pixel data representing a plurality of colors into a binary series signal suitable for arithmetic encoding. 6 is an arithmetic encoding unit which encodes the binary sequence signal by dynamic arithmetic encoding. Reference numeral 7 denotes a color character removal unit, which replaces the data of a pixel determined to be a color character with the average value data of the block to which the pixel belongs. 8 is an orthogonal transform unit, which performs DCT (Di 5crete
Co51ne Transform) and further Huffman encoding, so-called ADCT encoding. Reference numeral 9 denotes a code data transmitter, which integrates the outputs of the arithmetic encoder 6 and the orthogonal transformer 8 to generate code data to be transmitted.

10 is a code data receiving unit which separates the received code data into an arithmetic code and a Huffman code. Reference numeral 11 denotes an inverse arithmetic encoding unit, which decodes the arithmetic code and outputs color character data. 12 is an inverse orthogonal transform unit that performs Huffman decoding and inverse orthogonal transform, and outputs multivalued image data. A smoothing unit 13 performs smoothing to remove block distortion of the decoded image. Reference numeral 14 denotes a synthesizing section which synthesizes color characters and multivalued image data and outputs image data to be reproduced.

Reference numeral 15 denotes an image output section, which forms a visible image from the image data.

Each part will be explained below.

<Edge detection unit 2> In the edge detection unit 2, the second
As shown in the figure, the following calculations are performed between surrounding pixels A, B, C, and D to calculate the distance between two points in the RGB space and detect edges in the image. That is, the image data of the current pixel of interest and surrounding pixels are respectively (Xr,
g, Xb), (Ar, Ag, Ab), then S= ((Xr-Ar) 2+ (Xg-Ag) 2(
Xb-Ab) 2)”” ... (1) S>THI
(=100) ...... When (2) is satisfied,
It is determined that there is an edge between X and A.

Similarly, the presence or absence of an edge is determined between B, C, and D, and if any one of A, B, C, and D is determined to be an edge, the pixel of interest X is determined to be an edge. do.

In this way, the presence or absence of an edge is determined by calculating the distance in the three-dimensional color space between the pixel of interest and the surrounding pixels, so for example, it is possible to determine the presence or absence of an edge with the same brightness but with different hues or color sources. I can do it.

Therefore, the present invention is extremely effective in detecting colored characters.

In addition to the edge determination for each pixel, the color pattern and its determination signal, which will be described later, are also output.

Note that the method of selecting peripheral pixels is not limited to the above-mentioned example, and for example, eight peripheral pixels may be selected.

Alternatively, for example, the average value of the image data of A, B, C, and D may be calculated, and the above calculation may be performed between the average value and the pixel X.

<Color Detection Unit 3> The color detection unit 3 detects a plurality of predetermined limited colors using the following equation. Now, the image data of the pixel of interest X (
r, g, b), then r, g b < thl and r-g, g-b, br < th3
・ ・ ・ When (3), the pixel of interest X is determined to be K (black).

Similarly, r>th2 and g.

When b<thl and g-b <th3, X=R (red), when g>th2 and r b<thl and r-b<th3, X=C (green), b>th2 and r g< When th 1 and r-g<th3, X=B (blue), when r, g>th2 and b<thl and r-g<th3, X=Y (yellow r b>th2 and), g< When thl and r-b<th3, X=M (magenta), g+ b>th2 and r<thl and g-b
<th3 . . . (9) When X=C (cyan), color detection is performed.

Note that here, thl, th2. th3 is a predetermined threshold value, for example, thl=50. th2=205. , th3=30, the detection result is good.

The color detection signal is represented by three bits (R, G, B), and the correspondence between each detected color and the R, G, and B values is as shown in FIG.

<Color character determination unit 4> The color character determination unit 4 is a pixel in a block where the edge detection unit 2 has determined that there is a pixel corresponding to an edge, and the color detection unit 3 calculates the pixels according to the above equations (3) to (3). A pixel that satisfies any of the conditions 9) is determined to be a color character.

Binary Series Conversion Unit 5> The binary series conversion unit 5 converts eight color signals represented by 3-bit color character determination signals into binary sequence signals shown in FIG.

A block diagram of the binary sequence converter 5 is shown in FIG. Input data 200 to 202 are converted for each pixel by a conversion table 91 comprised of a ROM or the like into a maximum 7-bit signal 212 shown in FIG. 3, and input to a signal output device 92. Here, a plurality of ROMs may be prepared in the conversion table 91 so that short bit lengths can be assigned to colors that appear frequently, and the plurality of ROMs may be selected according to the control signal 300.

The signal output device 92 has a shift register configuration,
A 7-bit input signal 212 is input in parallel and serially output bit by bit starting from the MSB. This is the binary sequence signal D203. When the binary series signal becomes 1 or when the signal output device 92 outputs 7 O's, the signal output device 92 finishes outputting the color signal of one pixel and receives the next input data. In addition, the bit currently output from the signal output device 92
A signal Bt204 representing the number of bits in the binary sequence signal is output.

In this way, by converting the 3-bit color character determination signal into a binary sequence and encoding it as a 1-bit serial signal, it is possible to eliminate the need to separately encode the 3-bit signals that are correlated with each other. Encoding can be performed while preserving color correlation. Furthermore, when encoding is performed while predicting the pixel of interest, as in arithmetic encoding, the prediction and code are used as color information without predicting and encoding each color component of R, G, and B. , encoding efficiency can be increased.

In addition, since each color component of R, G, and B representing the color of each pixel is represented as one data, by decoding one data at the time of decoding, the R, G, and B signals corresponding to each pixel can be color images can be obtained, and color images can be reproduced quickly.

<Arithmetic Encoding Unit 6> In the arithmetic encoding unit 6, binary sequence signals representing several colors are encoded by arithmetic encoding, which is reversible encoding. The arithmetic encoding method and circuit configuration are as shown in Japanese Unexamined Patent Publication No. 2-65372.

Dark color character removal unit 7> The color character removal unit 7 processes the data of the pixel determined to be a color character by the color character determination unit 4i using a value corresponding to the data of other pixels in the block to which the pixel belongs. Replace.

That is, color character data is removed as shown in FIG. 5(b) from an image in which colored characters exist as shown in FIGS. 5(a, 5). In addition, at that time, by subtracting the colored characters, Figure 5 (b
) In order to remove the edges that occur as shown in Figure 5(c), we also subtract the data of pixels surrounding the color text and having the same hue as the color text pixels, and remove the data from other pixels in the block. Replace with the average value of pixel data.

At this time, it is assumed that the size of the block for color character removal processing and the block for orthogonal transformation described later are the same.

The configuration of the color character removing section 7 is shown in FIG.

Each of the 8-bit pixel image data r, g, and b is input to a color detector 71, which detects pixels of the color to be removed based on the above equations (3) to (9). At this time, in order to ensure that the peripheral parts of the colored characters are also detected, the threshold value is determined as shown below, for example.

thl=1.20 th2=130 th3=30
In this way, by changing the color detection threshold described above, the color detection unit 3
By performing color detection in a wider range, it is possible to extract a portion with a color tone similar to that of a colored character, and the input image data of this portion can also be removed.

When at least one of the detection signals R', G', and B' of the color detector 71 is 1, it is determined that the pixel has a color to be removed, and the subtraction circuit 72 calculates r, g, and b of the pixel.
Let the value of be O. Next, in the average value calculation circuit 73, 8
The average value of the r, g, b data in the X8 pixel block is calculated, and the replacement unit 74 replaces the average value as image data of the pixel from which the color has been removed, r', g', b'
Output as .

Note that the replacement is not limited to the average value, but can also be replaced with the most frequent value, or replaced with the median value of pixels within the block using a median filter.

In addition, in order to more accurately extract only the pixel surrounding the color character that is true and approximate to the color character, the signal obtained by ORing the color character determination signals R, G, and B and the signal 2 in FIG. One O
The output signals of the R circuit may be ANDed to perform processing in the subtraction circuit 72 and the substitution unit 74.

<Orthogonal transform unit 8> The orthogonal transform unit 8 performs a two-dimensional discrete cosine transform in units of 8×8 pixel blocks, performs quantum conversion on the obtained transform coefficients, and then performs Huffman encoding, which is the so-called ADCT encoding. .

FIG. 7 shows the configuration of the orthogonal transform section 8. In the preprocessing section 81, 8-bit signals of r', g', and b' are converted into a luminance signal Y and a chromaticity signal CrCb for each pixel. Next, in the sub-sampling section 82, Cr, Cb
An average value is taken for each double signal 2×2 pixel block. This takes advantage of the fact that the deterioration of the chromaticity signal is less noticeable to human vision than the deterioration of the luminance signal.Finally, in the orthogonal transform section 83, each of Y, Cr, and Cb is ADCT encoding is performed independently for each surface.

This encoding can be performed by configuring a dedicated arithmetic circuit or by computer software.

<Code Data Transmitter 9> In the code data transmitter 9, the color character pattern code is first transmitted, and then the Y, Cr, and Cb code data are transmitted in sequence. Prior to transmitting each side, a flag indicating which component the data is is transmitted.

At this time, it has a memory for compensating for time lag depending on the transmission order of each data.

As described above, by encoding color character patterns using reversible encoding, highly efficient data compression can be performed while maintaining the quality of color characters.

On the other hand, when separating color characters from original data, by performing predetermined replacement including surrounding parts, it is possible to improve the efficiency of orthogonal transform encoding.

Code data receiving section 10> The code data receiving section 10 receives the code data from the transmitting section 9 and determines whether it is an arithmetic code based on the flag, Y, Cr,
It is determined whether there is any Huffman code 51 of Cb, and the respective data is output to the inverse arithmetic encoding section 11 and the inverse orthogonal transform encoding section 12.

<Inverse arithmetic encoding unit 11, inverse orthogonal transformation encoding unit 12> The inverse arithmetic encoding unit 11 and inverse orthogonal transformation encoding unit 12 process color character data,
and ``+g'+b' multivalued data is decoded.

The decoded multivalued data of r', ', b' is
In the smoothing section 13, each surface is smoothed.

The reason why smoothing is not performed on color character data but only on multivalued data is to prevent deterioration of the resolution of color characters and reproduce clear color characters.

<Composition unit 14> The composition unit 14 combines the decoded color character data and r, g.
", b° multi-value data are synthesized.

That is, the result of multiplying color character data (R, G, B) by a predetermined coefficient a (RXa, GXa, BXa) and multi-value data (
r', ', b'). In this synthesis, priority is given to color character data as image data of pixels where color characters exist. This allows color characters to be reproduced with ease.

<Image output section 15> The image output section 15 is, for example, a laser beam printer, L
Image output devices such as ED printers, liquid crystal printers, thermal transfer blinks, dot blinks, inkjet printers,
An image display device such as a CRT, which forms a visible image on a recording medium in response to a reproduction signal.

In particular, inkjet printers include so-called bubble jet printers that use a type of head that ejects droplets by film boiling using thermal energy.

1. Example of the "Results" The present invention is not limited to image communication devices such as color facsimiles, but can also be applied to storage devices such as image files.

FIG. 8 shows an example in which the present invention is applied to a storage device.

In FIG. 8, 1 to 15 are the same as in FIG. 1, so their explanation will be omitted. 21 is hard disk, ROM, RA
This is an image file composed of files such as M, and can store multiple images. At the time of storage, the arithmetic code and Huffman code may be stored separately or may be stored together for each image. Also, for example, if you want to use only the text part on a display or hard copy,
It is only necessary to decode the arithmetic code (in that case, the processing time can be shortened).

Summer I Oni IL In this embodiment, in addition to the configuration of the first embodiment, a smoothing section 31 is further provided after the image composition section 14 at the time of decoding.

For smoothing, for example, a 3×3 pixel block smoothing filter can be used. Furthermore, as the filter coefficients, coefficients that take a weighted average of the pixel of interest and surrounding pixels may be used.

According to this embodiment, since smoothing is performed even after the color text and the multi-value image are combined, it is possible to prevent the boundary between the color text portion and the multi-value image from becoming unnatural. This is particularly effective when a COD sensor reads an original containing a mixture of text and natural images. Therefore, for example, in the case of an image such as computer graphics in which characters can be clearly separated, this smoothing may not be performed.

As described above, according to the embodiment of the present invention, the color character portions present in the input image are simultaneously detected and encoded simultaneously, so that multiple color characters can be encoded quickly. Moreover, by encoding separately from the gradation image part, highly efficient encoding can be performed while maintaining high quality. In other words, irreversible high-efficiency encoding is performed on gradation images, and in order to compensate for the drawback that high frequency components are lost due to the encoding of gradation images, edge parts, especially color character parts, are By entropy encoding, ringing can be prevented and color text parts can be reproduced with high quality.

Further, in addition to the color text portion, the surrounding color portions having approximately the same hue as the color text portion are also removed from the gradation image and predetermined replacement is performed, thereby greatly improving the efficiency of the gradation image.

Note that the image input section 1 is not limited to a CCD line sensor, but may be an interface that outputs a computer processing result, a still video camera that records still images, a video camera that records moving images, or the like.

In particular, computer interfaces include, for example, interpreters for page description languages such as BostScript and PCL.

In addition, input signals are not limited to R, G, and B color components; for example,
(Y, I, Q), (L″ a.

It may be a signal such as ``b''), (L'', u'', v''), (Y, M, C), etc.

Similarly, the color component signals for color detection are the same as those for R and G.
, is not limited to the B signal.

In addition, the encoding method of the above-mentioned color characters is not limited to binary series conversion and arithmetic encoding, but also run-length encoding, MH, MR, MM
Other reversible encoding such as R may also be used.

Further, the method of encoding a multivalued image is not limited to ADCT, but may be vector quantization or other orthogonal transform encoding.

Furthermore, the types and number of color characters to be detected are not limited to the above example.

Further, the circuit for extracting black characters of the above-mentioned prior art may be provided for multiple colors so that characters of multiple colors can be extracted.

Further, the present invention is not limited to an encoding device, but also an image processing device,
In particular, it is applicable to copying machines, color image editing devices, etc. that perform color conversion processing and line drawing extraction processing.

[Effects of the Invention] As described above, according to the present invention, it is possible to provide an image encoding device that can maintain good image quality and has high encoding efficiency.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the encoding device according to the first embodiment of the present invention. FIG. 2 is a diagram explaining edge detection. FIG. 3 is a diagram explaining binary sequence conversion. is a block diagram showing the configuration of the binary sequence conversion section, FIG. 5 is a diagram explaining color character removal, FIG. 6 is a block diagram showing the configuration of the color character removal section, and FIG. 7 is orthogonal transformation. FIG. 8 is a block diagram showing the structure of a second embodiment of the present invention. FIG. 9 is a block diagram showing the structure of a third embodiment of the present invention. 2...Edge judgment section 3...Color detection section, color character judgment section, binary series conversion section, color character removal section

Claims (3)

[Claims] (1) Extraction means for extracting a multi-color line drawing part of input image data; and encoding means for performing different encoding on the multi-color line drawing part extracted by the extraction means and parts other than the line drawing part. An image encoding device comprising: (2) The image encoding apparatus according to claim 1, wherein the extraction means extracts the line drawing portion of the plurality of colors by edge determination and color component determination. (3) The encoding means performs encoding after performing binary sequence conversion on the line drawing portion.
The image encoding device described in Section 2.