JPH0583569A - Image encoder - Google Patents

Abstract

PURPOSE:To reduce the redundancy of data bits at the image encoder to compress images by discrete cosine transformation(DCT). CONSTITUTION:A blocking circuit 2 divides image data into blocks defining 8X8 data in spatial arrangement as one block, and a transformation coefficient is calculated by executing the DCT for the image data in the respective blocks at a DCT circuit 5. The output of the DCT circuit is temporarily preserved in a buffer memory 7, and the luminance change of the block receiving the DCT is detected by a luminance change detection circuit 13. When detecting the block close to DC information reducing the luminance change, the data transmission area of this block is unequivocally decided (2X2) and encoded data are constituted by adding a flag only for one bit. By decreasing the flags composed of more than several bits to decide the transmission areas of 8X8 blocks close to DC data, the redundancy of the data bits in this block can be reduced.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image coding apparatus for highly efficient coding of image data by discrete cosine transform (hereinafter referred to as DCT).

[0002]

2. Description of the Related Art When transmitting image data or recording it on a recording medium such as a magnetic tape, a DCT method is known as a high efficiency encoding for image compression. DC
The transform coding by T utilizes the correlation of image signals, transforms sample values (hereinafter referred to as image data) into mutually orthogonal axes to make the correlation between image data uncorrelated, and The base vectors are orthogonal to each other, the sum of the average signal power before conversion and the sum of the average power of the transform coefficients obtained by orthogonal transform are equal, and the degree of power concentration on the low-frequency component of the image signal It is attracting attention as an excellent orthogonal transform.

The DCT divides an image into image blocks consisting of N (n × n) pixels in both horizontal and vertical directions in a two-dimensional space arrangement, and the image data in the image block is converted into a cosine function (cos function). Orthogonal transformation (two-dimensional DC
T). Then, the transform coefficient (DCT coefficient) obtained by this DCT is encoded only by the component in which power is concentrated, so that the amount of information can be significantly reduced as a whole.

Transform coefficients obtained by DCT transforming n × n image data are C _ij (i = 0 to n−1, j = 0 to n−).
Expressed as 1), the conversion coefficient C ₀₀ corresponds to the DC component representing the average luminance value in the image block, and its power is usually considerably higher than other components. Therefore, when the DC component is roughly quantized, block distortion, which is a noise peculiar to the orthogonal transform coding that is visually perceived as a large image quality deterioration, is generated. Therefore, a large number of bits (for example, 8 bits) is included in the transform coefficient C _00. The above is assigned and equally quantized, and the conversion coefficient C _ij (excluding C ₀₀ ) of other components excluding the DC component (hereinafter, referred to as the AC component) has a visual spatial frequency that decreases in a high frequency range. By utilizing the characteristics, the higher the frequency component, the smaller the bit allocation number and the quantization.

Further, in the area corresponding to one image block of the conversion coefficient C _ij , the significant conversion coefficient C having a value other than zero is obtained.
Find the maximum row number and column number where _ij (≠ 0) exists,
Enclose the significant transformation coefficient in the rectangle determined by this row and column,
Zone coding is performed in which only the transform coefficients included in the enclosed area (hereinafter, referred to as a coding area) are quantized to improve coding efficiency.

In the transmission and recording of image data, the transform coefficient C _ij obtained by DCT of the image data is quantized as described above, and then Huffman coding or run length coding is performed for further compression. Variable length coding such as
A synchronization signal, a parity, a flag and the like are added to the obtained encoded data for transmission and recording.

An example of a conventional example of the image coding apparatus will be described below. In FIG. 3, image data formed by sampling a video signal and converting it into a digital signal is input to a terminal 1 of the image encoding device. In the figure, a blocking circuit 2 has a storage capacity for one frame, a memory 3 for storing image data, and an image block in which the image data from the memory 3 has n × n in a spatial arrangement as one block. And a blocker 4 for reading each processing unit consisting of a predetermined number of image blocks obtained by dividing one frame into a plurality of blocks.

Then, the blocking circuit 2 stores the image data supplied through the terminal 1 in the memory 3 for each frame, and the image data stored in the memory 3 is, for example, 8 × in a spatial arrangement. The image block is divided into eight image blocks each having one block, and read out for each processing unit including a predetermined number of image blocks, and the read data is supplied to the DCT circuit 5.

The DCT circuit 5 is, for example, a digital circuit.
A two-dimensional DCT circuit which is composed of a signal processor and is composed of a DCT that performs orthogonal transform in the vertical direction using a cosine function and a DCT that performs orthogonal transform in the horizontal direction,
The image data supplied from the blocking circuit 4 for each processing unit is subjected to two-dimensional DCT to calculate a conversion coefficient C _ij , and the conversion coefficient C _ij is supplied to the quantization circuit 6.

The quantization circuit 6 quantizes the conversion coefficient C _ij from the DCT circuit 5 for each processing unit and the conversion coefficient C _ij read from the buffer memory 7. Quantizers Q1, Q2, Q3 forming quantized data, and for performing zone coding,
A region determining circuit 8 for determining a region (coding region) to be quantized, and a transform coefficient included in the coding region determined by the region determining circuit 8 is quantized and then coded by a Huffman code. A code amount calculating circuit 9 for detecting the data amount of the image block and controlling the quantization width of the quantizer based on the data amount, and a Huffman table circuit for giving the Huffman code table to the code amount calculating circuit 9. 10
Under the control of the code calculation circuit 9, the quantizers Q1 to Q1
A selector 11 for selecting one of the outputs of Q3. The quantized data selected by the selector 11 is supplied to the encoding circuit 12.

By the way, in the image coding apparatus,
At the time of high-efficiency image compression by DCT, image data to be recorded and transmitted is added with a flag representing the attribute of each block in addition to the DCT coefficient. Especially, the flag representing the transmission area of the AC (alternating current) coefficient is from several bits
Dozens of bits are transmitted. When the image data in the block is close to DC (is flat), AC is applied after quantization.
Almost no coefficient is transmitted, and only the DC component and the flag indicating the attribute of the block are transmitted. After quantization, the DCT coefficient is divided into a portion where the DCT coefficient becomes 0 (zero) and a portion having a value, and after the transmission area is determined, the transmission area is represented by a flag of several bits. When there are almost no components other than the components, most of the data to be transmitted is in the state of flags and AC coefficient 0.
Such a form of transmission would use the same number of bits to represent the absence of data, which would result in redundancy in the data.

[0012]

After quantization, the DCT coefficient is divided into a portion where the DCT coefficient becomes 0 (zero) and a portion having a value,
After determining the transmission area, the transmission area is represented by a flag of several bits. In such a method, when there are almost no components other than the DC component, most of the data to be transmitted is in the state of the flag and AC coefficient 0. This type of transmission uses the same number of bits to indicate that there is no data, and thus has a problem in that the data is redundant. The present invention is to reduce the redundancy of data bits in the above cases.

[0013]

According to the present invention, a block having a small change in luminance or a change in chroma in image information, that is, a block close to DC information is uniquely determined as a data transmission area. By omitting the flag indicating the transmission area, the number of bits to be transmitted and the number of bits to be transmitted are reduced for each block, and the redundancy is reduced by allocating this amount to other blocks. ..

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS At the time of image compression by DCT, data to be recorded and transmitted is added with a DCT coefficient and a flag representing an attribute of each block. Particularly, the flag representing the transmission area of the AC coefficient is transmitted from several bits to several tens of bits depending on the algorithm. When the image data in the block is close to (direct on) DC, the AC coefficient is hardly transmitted after quantization, and only the DC component and the flag indicating the attribute of the block are sent. When such a case is detected using the DCT coefficient, and when it is determined that the amplitude of components other than direct current is small, a flag indicating this is set to 1
And 0 in other cases. When this flag is 1, the flag indicating the transmission area is not transmitted, and only the data in the predetermined area is transmitted with the flag 1 added.

The technical idea of the present invention will be sequentially described with reference to FIGS. FIG. 2A shows an image block having a size of 8 × 8. In the block, a transmission region in which data exists and a non-transmission region in which data does not exist are horizontal column numbers H and vertical. It is divided at the position of the direction line number V. As an example of the encoded data in such a case, when the flags indicating the positions of H and V in the transmission area in the figure are 3 bits each and only the 2 × 2 area from the upper left corner is transmitted, conventionally, (010 010 DC coefficient AC coefficient) is formed. In this encoded data, the first 3 bits (010) represent the flag of the H position of the transmission area, the next 3 bits (010) represent the flag of the V position of the transmission area, and the DC coefficient Is 10 bits, AC
The coefficient is composed of a variable length code (VLC). In the case of such a code configuration, the flag indicating the transmission area has 6 bits.

Next, as another example of encoded data, as shown in FIG. 2B, an 8 × 8 block is divided into 2 × 2 small blocks and the presence or absence of data is transmitted. If there is data 1 and flag 0 if there is no data, and if a 2 × 2 area is transmitted from the upper left corner as in the above case, the encoded data of (1000..0 DC coefficient AC coefficient) Constitute. In the encoded data, the first 16 bits represent a data presence / absence flag, and DC
The coefficient is 10 bits, and the AC coefficient is a variable length code (VLC).

The present invention seeks to reduce the redundancy of such encoded data when the image data is as follows. If it is determined that there is no AC component in the 8 × 8 block or if it is smaller than a predetermined threshold value, as shown in FIG. 2C, a hatched area, for example, a 2 × 2 block is set as the data transmission area. By setting it uniquely,
The flag indicating the data transmission area is omitted. In such a case, the encoded data is composed of (1 DC coefficient AC coefficient), and 1 represents a flag when it is determined that there is no alternating current (AC) component or it is smaller than a predetermined threshold value,
DC coefficient is 10 bits, AC coefficient is variable length code (VL
C). When the flag is 1, the flag indicating the transmission area is not transmitted, but the data in the uniquely determined area is transmitted. In the opposite case, set the flag to 0.
To When the coded data is configured as described above, only 1 bit is added to the coded data configured in (A) of FIG.

The presence / absence of the AC component is determined as follows. The amplitude of the AC component of the 8 × 8 block can be detected using a threshold for each component of this block or a threshold for the sum of absolute values of all AC coefficients.

FIG. 1 shows an embodiment of the present invention, in which the presence or absence of the AC component is detected by detecting a change in luminance or a change in chroma using only the AC coefficient of DCT. Figure 1
In the circuit shown in, the image data input unit 1, the memory 3,
The blocker 4, the DCT circuit 5, the buffer memory 7, the quantizer Q, and the encoding circuit 12 are the same as those in the conventional example shown in FIG. 3, so the configuration characteristic of the present invention will be described below. In FIG. 1, 8 × 8 is provided on the output side of the DCT circuit 5.
A block luminance change detection circuit 13 is provided, and when the luminance change detection circuit 13 does not detect a luminance change, only one bit of the above-mentioned flag is added and transmission is performed as shown in FIG. The area is uniquely determined. After this process, the quantizer Q is controlled by the calculation and transmission area determination circuit 14 of the quantization code amount, and the quantized data is supplied to the encoding circuit 12. As another embodiment, the same object can be achieved by providing a circuit for detecting a change in chroma instead of the brightness change detection circuit 13.

[0020]

INDUSTRIAL APPLICABILITY According to the present invention, in the image information, the brightness change,
For blocks with a small change in chroma, that is, blocks close to DC information, the data transmission area is uniquely determined, and the flag indicating the transmission area is omitted. Can be reduced, and this amount can be allocated to other blocks. As a result, it is possible to reduce the compression rate of the block having a large number of AC coefficients of the DCT coefficient and reduce the quantization distortion.

[Brief description of drawings]

FIG. 1 is a block diagram showing a circuit configuration of an image encoding device to which the present invention has been applied.

FIG. 2 is an explanatory diagram of a transmission area of an image block.

FIG. 3 is a block diagram showing a circuit configuration of a conventional example.

[Explanation of symbols]

2 ... Blocking circuit 4 ... Blocking device 5 ... D
CT circuit 7 ... Buffer memory Q ... Quantizer 1
2 ... Encoding circuit 13 ... Luminance change detection circuit 14 ... Code amount calculation Transmission area determination circuit

Claims (1)

[Claims] 1. A block forming means for dividing image data into blocks each of which has n × n in a spatial arrangement, and image data of each block from the block forming means is orthogonally transformed by using a cosine function. Discrete cosine transform means for calculating transform coefficients, quantizing means for quantizing transform coefficients from the discrete cosine transform means to form quantized data, and outputting the quantized data; An image code comprising: a unit that detects a block close to DC information based on a conversion coefficient and uniquely determines a data transmission area of the block; and a unit that encodes an output of the quantization unit. Device.