JPH0795422A - Picture data compressing device - Google Patents

Abstract

PURPOSE:To improve the compression efficiency by operating the least significant bit of additional bits to break the continuity of bits '1' as the cause of byte stuffing and reducing the number of stuffing bytes. CONSTITUTION:Additional bits and the code size are generated in an additional bit encoding circuit 3. Data is encoded to a Huffman code in a Huffman code encoding circuit 4 by the code size and the zero run. Additional bits as the output of the additional bit encoding circuit 3 are inputted to an LSB control circuit 8. In the LSB control circuit 8, LSB (least significant bit) of additional bits is operated to break the continuity of bits '1' as the cause of byte stuffing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image data compression apparatus for editing image data on a personal computer or the like.

[0002]

2. Description of the Related Art In recent years, it has become possible to edit pictures such as pictures and paintings by inputting them as image data into a personal computer by using an image scanner or the like and making corrections and processing. However, as the number of edited image data increases, the capacity of the storage device required to store the image data also increases, and it is necessary to compress the image data for data communication and data exchange with other devices. Came out. Current,
As a highly efficient compression method for still images, JPEG (Joint
Photographic Expert Group) block coding
It is standardized in ISO-IEC-CD-10918-2.

FIG. 6 is a block diagram of a conventional JPEG block encoding apparatus. The image data converted into a block of horizontal 8 pixels and vertical 8 pixels by performing DCT (discrete cosine transform) processing is input to the zigzag scan circuit 61. The zigzag scanning circuit 61 performs zigzag scanning and rearranges in one dimension. The output of the zigzag scan circuit 61 is the zero-run counter 62.
To the additional bit encoding circuit 63.

The run length, which is the length of consecutive zero coefficients in the zero run counter 62, is assigned. Additional bits and codes in the additional bit encoding circuit 63
Create size.

In the Huffman coding circuit 64, Huffman coding is performed according to the code size and zero run. In the multiplexor circuit 65, the additional bit encoding circuit 6
Huffman coding circuit 64 and the additional bit which is the output of 3
The Huffman code, which is the output of, is fed into a single data stream.

In the word wrap circuit 66, the output of the multiplexer circuit 65 is divided into 8-bit byte units and output to the byte stuffing circuit 67. In the byte stuffing circuit 67, data in which all 8 bits are 0 is inserted on the data code side. Part-Time Job·
In the stuffing circuit 67, the byte stuffed data is recorded on a recording medium such as a hard disk.

In this conventional example, a marker code is used to add information necessary for decompression such as the size of the screen and the compression method. At the time of compression, a marker code is added to the compressed image data, and decompression is performed based on this.

The marker code is "0xFF" + "0x
It is represented by a 2-byte code of "00". Here, if the same pattern as the marker code appears in the compressed image data, inconvenience occurs. Therefore, in order to identify this, "0x00" of the compressed image data is detected at the time of compression. Immediately after the byte, add the stuffing byte "0x00".

At the time of decompression, "0xF" in the compressed image data
The byte "0x00" next to the byte "F" is skipped. The probability that the same pattern as the marker code will occur in the image data at the time of compression is
It was about 0.1 to 1.2% in one image data. In an artificially created / edited image such as computer graphics (CG) or a simple inset screen, the frequency of occurrence of stuffing bytes tends to increase as compared with a general landscape image. This stuffing byte was the cause of the reduction in compression rate.

[0010]

As described above, according to the conventional image data compression apparatus, the stuffing byte is added to the data code side in order to distinguish the marker code from the data code. It was difficult to improve the compression efficiency.

Therefore, the present invention eliminates the above-mentioned drawbacks and operates the LSB (least significant bit) of the additional bit to change the byte.
The purpose is to improve the compression efficiency by cutting off the continuation of the bit "1" which is the cause of the stuffing and reducing the stuffing byte.

[0012]

In order to achieve the above object, according to the present invention, a zigzag scanning means for rearranging the input first image data in zigzag order and the zigzag scanning means are provided. Additional bit encoding means for creating additional bits and code size from the second image data, zero run counter converting means for assigning run length of the second image data, the code size and the run length Huffman coding means for Huffman coding by, a multiplexer means for sending the Huffman code and the additional bits output from the Huffman coding means into a single data stream, and the data stream in byte units. The word wrap means for splitting and the least significant bit that operates the least significant bit of the additional bits. And a byte stuffing processing means for byte stuffing the code divided by the word wrapping means, and a means for suppressing the occurrence frequency of stuffing bytes by the least significant bit control means. .

[0013]

In the above-mentioned structure, the least significant bit control means manipulates the least significant bit of the additional bits to break the continuation of the bit "1" which is the cause of byte stuffing. It is possible to improve the compression efficiency.

[0014]

Embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of the present invention. The image data converted into a block of 8 horizontal pixels and 8 vertical pixels by DCT (discrete cosine transform) processing is
Input to the zigzag scan circuit 1. The zigzag scanning circuit 1 performs zigzag scanning and rearranges in one dimension. The output of the zigzag scan circuit 1 is input to the zero-run counter 2 and the additional bit encoding circuit 3.

Successive zeros in the zero run counter 2
Assign a run length that is the length of the coefficient of. The additional bit encoding circuit 3 creates an additional bit and a code size.

In the Huffman encoding circuit 4, the code
Huffman coding by size and zero run. FIG. 2 is an explanatory diagram of a Huffman code with additional bits added. In the figure, HF indicates a Huffman code, and AB indicates an additional bit.
The additional bit output from the additional bit encoding circuit 3 is L
Input to the SB control circuit 8. In the LSB control circuit 8, the LSB (least significant bit) of the additional bit is operated to cut off the continuation of the bit "1" which is the cause of byte stuffing at the AB end of 21 in FIG.

In order to perform this processing, the additional bit L
Replace SB with "0". Here, if the additional bit is positive, that is, the most significant bit (MSB) of the additional bit is "
The operation of the LSB is different between the case of 1 ”and the case where the additional bit is negative, that is, the case where the MSB of the additional bit is“ 0 ”.

In this embodiment, when the additional bit is positive as shown in FIG. 3, if LSB = 0, no change is made and the LS is changed.
If B = 1, 1 is subtracted to set LSB = 0. If the additional bit is negative, no change is made if LSB = 0, and LSB =
If it is 1, 1 is added to set LSB = 0. Here, if the additional bit is negative and LSB = 1 as shown in FIG.
Although there is also a method of reducing, the absolute value of the data will increase compared to before the LSB processing, and the quantization coefficient will increase,
Since a data overflow may occur during decryption, the operation of adding 1 is performed.

In the multiplexer / plexer circuit 5, the LSB is
The additional bits output from the control circuit 8 and the Huffman code output from the Huffman encoding circuit 4 are combined into a single data
Send to stream.

In the word wrap circuit 6, the output of the multiplexer circuit 5 is divided into 8-bit byte units,
Output to the byte stuffing circuit 7. In the byte stuffing circuit 7, data in which all 8 bits are 0 is inserted on the data code side. The data byte-stuffed in the byte-stuffing circuit 7 is recorded on a recording medium such as a hard disk.

In this embodiment, the LSB control circuit 8
By performing the above-mentioned LSB operation in 1), the occurrence frequency of the data code “0xFF” can be suppressed. As a result, the number of stuffing bytes is reduced, and the amount of compressed image data can be reduced.

Next, another embodiment of the present invention will be described. This embodiment is a modification of the operation of the LSB control circuit 8 in the embodiment shown in FIG. In this embodiment, when the additional bit is positive as shown in FIG. 5, LSB =
If 0, no change is made, LSB = 1, and LSB is 0
If byte stuffing can be avoided by changing to, subtract 1 from the added bit. Further, if LSB = 1 and byte stuffing cannot be avoided even if LSB is changed to 0, or if byte stuffing does not occur even when LSB = 1, the additional bit is not changed.

If the additional bit is negative, no change is made if LSB = 0 and LSB = 1, and 1 is added to the additional bit if byte stuffing can be avoided by changing LSB to 0. Also, LSB = 1, and LSB
If the byte stuffing cannot be avoided even if is changed to 0, or if the byte stuffing does not occur even if the LSB remains 1, the additional bit is not changed.

In this embodiment, since the LSB of the additional bit is manipulated only when the occurrence of byte stuffing can be reduced, it is possible to suppress the deterioration of the image and improve the compression efficiency.

[0025]

As described above, according to the present invention, it is possible to suppress the occurrence of byte stuffing and increase the compression efficiency by operating the LSB of the additional bit.

[Brief description of drawings]

FIG. 1 is a block diagram showing a data compression apparatus according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram of a Huffman code with additional bits added.

FIG. 3 is an explanatory diagram showing control of additional bits.

FIG. 4 is an explanatory diagram showing control of additional bits.

FIG. 5 is an explanatory diagram showing control of additional bits.

FIG. 6 is a block diagram showing a conventional image data compression device.

[Explanation of symbols]

1 ... Zigzag scan circuit, 2 ... Zero run counter, 3 ... Additional bit coding circuit, 4 ... Huffman coding circuit, 5 ... Multiplexer circuit, 6 ... Word wrap circuit, 7 ... Byte stuffing circuit, 8 ... LSB control circuit.

Claims (1)

[Claims] 1. A zigzag scanning means for rearranging the input first image data in a zigzag order, and an addition for creating additional bits and code size from the second image data obtained by the zigzag scanning means. Bit encoding means, zero-run counter forming means for allocating the run length of the second image data, Huffman encoding means for performing Huffman encoding with the code size and the run length, and the Huffman encoding means Huffman code which is the output of the above and the additional bit which sends the additional bit into a single data stream, word wrapping means which divides the data stream into bytes, and the least significant bit in the additional bit Divided by the least significant bit control means for operating And byte stuffing processing means a byte-stuffing code was, the image data compression apparatus characterized by comprising a means to reduce the frequency of occurrence of stuffing bytes by the least significant bit control means.