JPH0345083A - Picture encoder - Google Patents

Abstract

PURPOSE:To attain encoding with one reciprocating scanning by providing a decentralizing processing means applying decentralizing processing based on the scanning of an original at a forward motion and coding processing means applying encoding based on the original scanning at return. CONSTITUTION:A switch 5 selects a decentralizing processing section 6 at a forward path of a scanning optical system 30 to execute the decentralizing processing, the decentralizing processing section 6 generates a bit assignment table based on the result of decentralizing processing and the result is stored in a bit assignment table storage section 8. A switch 9 is thrown to the position of a transmission line 10 and the content of a bit assignment table of the bit assignment table storage section 8 is sent to a transmission line 10 as header information. On the other hand, the switches 5, 9 select a quantization section 7 in the return path of the scanning optical system 30, a data is read and a 2-dimension DCT conversion section 3 applies 2-dimension DCT conversion to apply NXN coefficient from an NXN buffer 4 and a bit assignment table read from the bit assignment table storage section 8 to the quantization section 7, in which the quantizing processing is executed. Thus, encoding is attained by reciprocating the original scanning once.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to an image encoding device for compressing image data obtained by scanning a document.

[Conventional technology]

It is known that an encoding method using cosine (COS) transformation is effective when compressing digitized gradation image data. In this transform encoding method, the image is N
The image is divided into a plurality of blocks of XN pixels, and each block is subjected to two-dimensional discrete cosine transform (hereinafter referred to as two-dimensional DCT transform) to obtain transform coefficients, and each component is quantized. At this time, the number of quantization levels for each component is determined using the variance of each component.

When encoding an image read by a scanner as described above, it is necessary to read out the image information scanned once twice. In the first reading, after performing two-dimensional DCT transformation, the statistics of each component are taken to find the variance, and the number of quantization levels for each component is determined. Then, in the second readout, the readout image is subjected to two-dimensional DCT transformation, and then encoded at the quantization level determined as described above.

Note that, for example, Japanese Patent Application Laid-Open No. 109662/1983 is related to this type of device.

[Problem to be solved by the invention]

However, in the above-mentioned conventional technology, since uncompressed image information is stored, a large-capacity storage device is required, resulting in an increase in cost. Therefore,
The document is scanned twice, and in the first scan, after performing two-dimensional DCT transformation, the statistics of each component are taken to find the variance, and the number of quantization levels for each component is determined. In the second scan, the read image is subjected to two-dimensional DCT transformation, and then encoded at the quantization level determined as described above.

However, although this method does not require a storage device to store uncompressed images, it does require the image reading unit to move back and forth over the document twice to scan, which reduces work efficiency. There's a problem.

SUMMARY OF THE INVENTION An object of the present invention has been made in view of the above-mentioned state of the prior art, and it is an object of the present invention to provide an image encoding device that can perform compression processing without having to reciprocate scanning of an image reading unit twice.

[Means for Solving the Problem] In order to achieve the above object, the present invention optically reads an image of the document with a scanning optical system that moves relative to the document,
In an image encoding device that performs compression processing by performing two-dimensional discrete cosine transformation on the read information, there is provided a distributed processing means that performs distributed processing based on statistical properties of the read information obtained by scanning the document during the forward trip; and encoding processing means for encoding based on the information read by scanning the original and the processing results of the distributed processing means.

[Effect]

According to the above means, a two-dimensional discrete cosine transform is performed on the image information read in the forward path of document scanning, the variance is determined from the statistics of each component, and from this variance value, the quantum The number of conversion levels is determined. next,
A two-dimensional discrete cosine transform is similarly performed on the image information read during the return pass of document scanning, and the result is encoded based on the number of quantization levels determined during the return pass. Therefore, it becomes possible to achieve encoding with one round trip scan.

〔Example〕

Hereinafter, the present invention will be explained in detail with reference to FIGS. 1 to 3.

FIG. 1 is a block diagram showing an embodiment of an image encoding device according to the present invention, FIG. 2 is a schematic front view showing details of a reading section, and FIG. 3 is a flowchart showing the processing of the present invention.

As shown in FIG. 1, an N line memory 1 is provided to store every N lines digital signals of image information of a document optically read by a reading section.
An NXN buffer 2 is connected to the line memory. The NXN buffer 2 is connected to a two-dimensional DCT transform unit 3 that performs the two-dimensional DCT transform described above.
An NXN buffer 4 is connected to the T converter 3.

A switch 5 that switches in conjunction with a switch 9 (to be described later) is connected to the NXN buffer 4, and a distributed processing section 6 that performs distributed processing is connected to one of its outputs, and a quantization section 7 that performs quantization processing is connected to the other output. It is connected. A bit allocation table storage section 8 is connected to the distributed processing section 6, and a switch 9 is connected to its output terminal. The switch 9 selectively outputs the output signal of the bit allocation table storage section 8 to the output terminal or the quantization section 7 .

In FIG. 2, a glass stand 21 on which a document 22 is placed is disposed on the upper surface of the apparatus, and a scanning optical system 30 forming a reading section is disposed below the glass stand 21. This scanning optical system 30 reciprocates the lower part of the glass table 2 in the main scanning direction to read the image of the document. The optical system is composed of a light source 23 that illuminates the original 22, a lens array 24 that focuses reflected light from the original surface on a predetermined position, and a photoelectric conversion unit 25 that converts light from this lens array 24 into electricity. There is.

The drive mechanism for reciprocating the scanning optical system 30 includes a wire 26 that is attached to the scanning optical system 30, pulleys 27a and 27b that rotatably load this wire 26, and this pulley 27a.
, 27. It is comprised of a motor 28 serving as a drive source for rotating the motor 28, a pulley 29 attached to the rotating shaft of the motor 28, and a belt 30 installed between the pulley 29 and the pulley 27b.

In such a configuration, when a scan start command is issued,
The motor 28 rotates, and the rotation is transmitted to the pulley 27b, causing the wire 26 to rotate clockwise. As the wire 26 rotates, the scanning optical system 30 moves horizontally in the direction B shown in the figure. When the scanning optical system 30 reaches the scanning end position, the motor 28 stops rotating and then reverses. This reversal causes the wire 26 to rotate counterclockwise and the scanning optical system 30 to move in the illustrated inward direction. Scanning optical system 30
In the process of movement, the photoelectric conversion unit 25 reads the image pixel by pixel in the sub-scanning direction and outputs the density of each pixel as a multi-value signal. This output signal is manually input to the N line memory 1 shown in FIG.

Next, the operation of the embodiment of the present invention will be explained with reference to FIG.

The multilevel signal output from the photoelectric conversion section 25 is stored in the N line memory 1 every N lines (step 31). N
The signal stored in the line memory 1 is read out to the NXN buffer 2 at a predetermined timing and inputted to the two-dimensional DCT conversion section 3.

The two-dimensional DCT transformation unit 3 divides the image information into N×N blocks, and sequentially performs orthogonal transformation on each block by two-dimensional DCT transformation (step 32). Here, the image signals are Xo, 1. Two-dimensional DCT transformation in the case of j=0 to N-1 is defined by the following equation.

(However, u, ■=0, 1..., N-11C(w)
= 1/aT:, w=OC(w)=1, w=0
．． 1.2 to N) The conversion output yUv from the two-dimensional DCT conversion unit 3 is stored in the NXN buffer 4.

During the forward path of the scanning optical system 30 (direction B in FIG. 2), the switch 5 selects the dispersion processing section 6, so dispersion processing is executed (step 33). The distributed processing unit 6 creates a bit allocation table based on the distributed processing results (step 34).
) and stores this in the bit allocation table storage section 8.

At this time, since the switch 9 has selected the transmission path IO side, the bit allocation table contents of the bit allocation table storage section 8 are sent to the transmission path 10 as header information.

On the other hand, on the return path of the scanning optical system 30, the switches 5 and 9 select the quantization section 7. As in the case of the outbound trip, the data is read (step 35) and subjected to two-dimensional DCT.
A two-dimensional DCT transformation is performed by the transformation unit 3 (step 36).

Next, the NXN coefficients from the NxN buffer 4 and the bit allocation table read from the bit allocation table storage section 8 are applied to the quantization section 7, and based on these, the quantization section 7 executes quantization processing ( Step 37).

The bit allocation for each coefficient is determined from the variance value of each component. Then, the variance value of each coefficient is expressed by the following equation.

(K is the total number of blocks) Furthermore, the number of bits allocated to each coefficient is expressed by the formula.

muv=+ (10g 2 σuv quadratic (where M is the number of bits allocated for l block) In this way, by quantizing coefficients with large variance with a large number of bits, and quantizing small coefficients with a small number of coefficients,
Even if the overall number of bits is reduced, it is possible to obtain an image with less visual deterioration when reproduced. Note that when restoring encoded data, the process opposite to that of encoding may be performed. That is, the coefficients are obtained by inverse quantization, and the image is reproduced by performing two-dimensional DCT inverse transformation. The quantized result is sent to the transmission line 10 (step 38).

Note that the receiving section (not shown) that receives the information sent from the quantizing section 7 to the transmission path 1O dequantizes the quantized value based on the bit allocation table sent first, and then dequantizes the quantized value. The image is reproduced by performing two-dimensional DCT inverse transformation.

As described above, image information can be compressed based on the statistical properties of the image by only scanning the original once back and forth.

〔Effect of the invention〕

As explained above, according to the present invention, an image code that optically reads an image of a document with a scanning optical system that moves relative to the document, performs two-dimensional discrete cosine transformation on the read information, and performs compression processing. The encoding device includes a distributed processing unit that performs distributed processing based on statistical properties on information read by scanning a document during the forward trip, and a distributed processing device that performs encoding based on the information read by scanning the document during the return trip and the processing result of the distributed processing device. Since an encoding processing means is provided, encoding can be achieved without scanning the document twice, and a page memory or the like can be made unnecessary.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of an image encoding device according to the present invention, FIG. 2 is a schematic front view showing details of a reading section, and FIG. 3 is a flowchart showing the processing of the present invention. l...N line memory, 2...NXN buffer, 3
...2-dimensional DCT conversion unit, 4...NXN buffer,
5, 9...Switch, 6...Distributed processing unit, 7...
- Quantization unit, 8...Bit allocation table storage unit, 10
...Transmission line, 30...Scanning optical system.

Claims (1)

[Claims] In an image encoding device that optically reads an image of the document with a scanning optical system that moves relative to the document, and performs a two-dimensional discrete cosine transform on the read information and performs compression processing, the image is read by scanning the document during the forward pass. It comprises a distributed processing means that performs distributed processing based on statistical properties obtained from the information, and an encoding processing means that performs encoding based on information read by scanning a document during the return trip and the processing result of the distributed processing means. An image encoding device characterized by: