JPH0267079A - Image encoder - Google Patents

Abstract

PURPOSE:To obtain satisfactory gradation data without losing a character or a thinning line by sampling multivalue data in which binary image data is reversely quantized, and applying a binarization processing on the data by an error diffusion method. CONSTITUTION:Binary data 100 representing an original image is stored in a first frame memory, and is converted to the multivalue data at a first reverse quantization circuit 2, and furthermore, the multivalue data sampled at a first sampling circuit 3 is binarized at a first binarization circuit 4 by the error diffusion method. Binary data 300, on which a low-pass filtering processing being applied again at a second reverse quantization circuit 5, is converted to the multivalue data, and furthermore, is sampled to the resolution of 1/2 at a second sampling circuit 6. The multivalue data is binarized at a second binarization circuit 7 by the error diffusion method, and is encoded at an encoder circuit 10, then, transmitted. In such a way, it is possible to perform satisfactory resolution conversion without losing the information of the original image.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image encoding device that hierarchically encodes a binary image.

[Prior Art] Conventionally, a method of hierarchically encoding a binarized image is to generate images with different resolutions by subsampling the image, and then sequentially encode these images starting from the lowest resolution. It was getting worse.

[Problems to be solved by the invention] However, since only subsampling is performed on image data, there is a problem that characters and thin lines are cut off or disappear before encoding. , coded data obtained by coding a low-resolution image does not represent the image properly, and an image obtained by decoding the root coded data cannot accurately reproduce the original image.

[Means for solving the problem (and operation)] The present invention has been made in view of the above points, and includes an inverse quantization means for converting binary image data into multi-value data, and a sub-quantization means for converting multi-value data into sub-value data. An image encoding device comprising a subsampling means for sampling, a binarization means for binarizing while diffusing errors generated during binarization to surrounding pixels, and an encoding means for encoding the binarized data. It provides:

This configuration prevents characters and thin lines from becoming cut off or disappearing before encoding, thereby achieving good encoding.

[Embodiment] FIG. 1 is a block diagram showing an embodiment of an encoder to which the present invention is applied.

First, the binary data 100 representing the original image is
It is stored in frame memory 1. Next, the binary data 200 read from the first frame memory 1 is subjected to low-pass filtering processing (3×3 filter) in the first dequantization circuit 2 and converted into multi-value data. Furthermore,
In the first sub-sampling circuit 3, the image is sub-sampled to an arbitrary resolution (here, 1/2 of the original image). The subsampled multilevel data is binarized by the first binarization circuit 4 using the error diffusion method.

The binary data 300 is processed by the second inverse quantization circuit 5 and the selector 8.
The second inverse quantization circuit 5 performs low-pass filtering processing (3×3 filter) again to convert the data into multi-valued data, and the second sub-sampling circuit 6 further converts the image into 1/2 resolution (original image). 1/4 for
) is subsampled. This multivalued data is converted into two values by the error diffusion method in the second binarization circuit 7 as described above.
Value conversion is performed.

The selector 8 selects the first
One of the signal 200 from the frame memory, the signal 300 from the first binarization circuit, and the second binarization circuit signal 400 is selected and output as a signal 500.

The signal 500 is subjected to, for example, MH encoding in the encoding circuit 10 and then transmitted.

On the receiving side, a decoder 11 decodes the MH code, and a display 12 displays the decoded data.

FIG. 2 is a block diagram of the first inverse quantization circuit 2 described above. The manually input binary signal 20 is held in latches 44a, b, and c with a delay of one pixel clock each. Also, line memory 43-a.

latches 44d, e, f, and latches 43g, h, i.
, signals whose pixel positions correspond to latches a, b, and c are obtained. As a result, binary data of 9 pixels forming the 3×3 matrix shown in FIG. 3 is obtained. The output signals from the latches 44a, c, g, and i are summed by an adder 45a, and multiplied by a constant (x
C+) operation is performed.

Further, the output signals from the latches 44b, d, f, and h are summed by an adder 45b, and multiplied by a constant (X
C2) Will be done. Further, the output signal from the latch 44e, which is the median value, is multiplied by a constant (XCs) by the multiplier 46c. The values of C1°C2, C3 indicate the degree of inverse quantization and can be set by the external controller 9, but the standard values are CI, C2, C3=1.

The output signals of the multipliers 46a, b, and c are summed by an adder 47, and then outputted as a multilevel signal 30. Furthermore, the second
The inverse quantization circuit 5 can be realized with a configuration completely similar to that shown in FIG.

Figure 4 shows the subsample circuit 3 used in the encoder shown in Figure 1.
．． 6 is an explanatory diagram of the operation of No. 6. By extracting the pixel data indicated by diagonal lines in the figure at every other timing in the main scanning and sub-scanning directions, 1/2 size (1/4 area)
A sub-sampled image is formed. This can be easily achieved by adjusting the latch timing of image data.

In this way, before the subsampling operation for reducing the resolution, binary data is converted into multi-value data by taking data of surrounding pixels into consideration using an inverse quantization circuit as shown in the second diagram. As a result, image data that does not correspond to the pixel position of the subsample is not ignored, but is taken into consideration during sampling processing, resulting in good resolution without losing information in the original image. Conversion can be achieved.

FIG. 5 shows the first binarization circuit 4 using the error diffusion method of the embodiment.
FIG. Multivalued image data Xlj from the first sub-sampling circuit 3 is input to the terminal 900 . 6
7 is a multiplier, which is standardized to a value of 255 or less on human power data. 60 is an error buffer memory, which stores density errors ε1.60 that have been diffused up to a certain point. is stored for each pixel. Error εIJ used in error buffer memory 60
The group follows the multi-value image data XIJ by moving the corresponding window 60 as the human-powered multi-value image data XIJ advances. 61 is a weighting coefficient circuit, which calculates the effective error ε8. is multiplied by a predetermined weighting coefficient αkl to form a standardized correction value.

Figure 6 shows an example of the arrangement of weighting coefficients α□.
Assuming that the pixel position in the error buffer memory 60 of the multivalued image data XIJ is 66, this position also corresponds to the * mark position of the weighting coefficient αkl.

62 is an adder which adds the standardized correction value to the manually generated multivalued image data XIJ to form correction data x'1''. The above processing is expressed by the following equation.

Σαkl’　ε1◆k　”◆1 In the above equation, the second term on the right side is a standardized correction value.

Next, the correction data
(for example, 128 gradations), and the binarization circuit 63 outputs logic "1" as the Yth output data if X'IJ≧T4.
” level (equivalent to 235 gradations) and
If not ≧T4, a logic "0" level (corresponding to 0 gradation) is output as output data YIJ. In this way, the multivalued correction data X'IJ is binarized. Binarized data YIJ
is stored in the output buffer 65.

64 is an arithmetic unit, which calculates correction data
' , , is subtracted, and the density error ε newly generated in the printing of the relevant pixel is calculated at the pixel position 66 of the error buffer memory 60.
to be recorded. By sequentially repeating this operation by advancing the input multivalued image data XIJ pixel by pixel, the binarization process using the error diffusion method is executed. Note that although error data of surrounding pixels is used here, it may be used for four surrounding pixels, and is not limited to the above embodiment. Furthermore, by raising the coefficients of the circuit to this power, the weighting circuit can be constructed using a shift operation. Further, the second binarization circuit 7 can also be realized with a configuration similar to that shown in FIG.

In this way, instead of simply binarizing the multivalued image data whose resolution has been reduced by being subsampled in the subsampling circuit 3 or 6, the system performs binarization using the error diffusion method. be.

As a result, errors that occur during the binarization of multi-value data are diffused to surrounding pixels, and the multi-value data is binarized while preserving the image density, which prevents information from being lost due to the binarization process. Therefore, characters, thin lines, etc. can be binarized well, and it is possible to prevent them from being binarized in a state where they are cut off or disappear.

The selector 8 selects the signal 2 according to the control of the controller 9.
One of 00, 300, and 400 is output as signal 500. For example, in the first step, signal 400 is selected and sent to encoding circuit 10. This makes it possible to encode an image with 1/4 the resolution of the original image. Once this 1/4 resolution image has been transmitted, the process moves to the next step.

The next step is to select signal 300 and encode and transmit the original 1/2 resolution image. After completing this, the signal 200 is selected, and the original image is encoded and transmitted.

As described above, images can be hierarchically encoded and transmitted.

Note that here, a general MH encoder/decoder is used as the encoding circuit 10 and the decoding circuit 11, so a description thereof will be omitted. Further, the encoder/decoder is not limited to the MH code/decoder, but an encoder/decoder for MR, MMR, arithmetic code, etc. may be used.

Furthermore, for color images, by preparing the circuits shown in this embodiment for predetermined colors, a hierarchical encoding device for color binary images with less deterioration of characters and thin lines can be obtained.

Further, in this embodiment, the number of layers is set to three, but the number is not limited to this, and by having a necessary number of inverse quantization circuits, sub-sampling circuits, and binarization circuits, an N-layer encoding device can be provided.

It goes without saying that the above-mentioned processing including this embodiment can be easily realized by software processing using a microprocessor.

[Effects of the Invention] As explained above, according to the present invention, by subsampling multivalued data obtained by dequantizing binary image data and performing binarization processing using the error diffusion method, the original image can be obtained. Since the image density is preserved, it is possible to obtain good hierarchical data (encoded data) in which characters and thin lines do not disappear.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of an embodiment of an encoder to which the present invention is applied, Fig. 2 is a block diagram of an inverse quantization circuit, Fig. 3 is a diagram showing filter coefficients of the inverse quantization circuit, and Fig. 4 is a sub- FIG. 5 is a block diagram of a binarization circuit; FIG. 6 is a diagram showing an example of weighting coefficients; 2 is a first inverse quantization circuit; is the first sub-sample circuit, 4 is the first binarization circuit, 5 is the second dequantization circuit, 6 is the second sub-sample circuit, 7 is the second binarization circuit, 8 is the selector, and 10 is the This is an encoding circuit.

Claims (1)

[Claims] An image encoding device that hierarchically encodes and transmits a binary image, comprising: inverse quantization means for converting binary image data into multi-value data; sub-sampling means for sub-sampling the multi-value data; An image encoding device comprising: binarization means for binarizing while diffusing errors generated during digitization to surrounding pixels; and encoding means for encoding the binarized data.