JP3517454B2 - Image encoding apparatus and image encoding method - Google Patents

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 and an image coding method, and more particularly to image coding using orthogonal transform coding.

[0002]

2. Description of the Related Art As a device using an image coding apparatus of this type, a digital VTR which digitizes an image signal and performs recording / reproduction has been conventionally known. In such a digital VTR, a compression coding technique is used for compressing image data and performing recording / reproduction. An orthogonal transform coding method is known as a technique for compressing and coding image data with high efficiency. This is a method in which image data is grouped into a plurality of pixels into blocks, and then the discrete cosine transform (discrete cosine transform) is performed.
Orthogonal transformation such as DCT) is performed, and the coefficient after the transformation is quantized, entropy coded, or the like.

FIG. 6 shows a schematic structure of a conventional image coding apparatus using DCT. In FIG.
Reference numeral 100 is input digital image data, 101 is a memory device for recording the image data 100, 103 is image data 102 stored in the memory device 101, read in block units of a predetermined size, and discrete cosine transform is performed for DCT.
A DCT device that outputs a transform coefficient (DCT coefficient) 105,
A buffer device 106 delays the DCT coefficient 105,
Reference numeral 108 is a quantizer for quantizing the delayed DCT coefficient, and 107 is an AC coefficient 104 of the DCT coefficient 105.
From the above, a definition determination device for determining the definition of the image data in the frequency domain, 109 is a quantization control device for controlling the quantization device 108 based on the definition of the image determined by the definition determination device 107, and 110 is In the encoding device, for example, run length encoding and two-dimensional Huffman encoding are used together to perform variable length encoding on the quantized data. Reference numeral 111 is an encoding control device for controlling the encoding device 110, and reference numeral 112 is variable-length encoded data.

Next, the operation will be described. The input image data 100 is stored in the memory device 101 in units of frames.
The pixels are divided into blocks of 8 pixels, and the blocks are shuffled and read. Memory device 1
The image data 102 read from 01 is the DCT device 1
03. The DCT device 103 performs DCT on the image data 102 for each 8 × 8 pixel block, converts the image data 102 from spatial domain to frequency domain data,
The DCT coefficient 105 that is the conversion coefficient is output. Figure 7
Shows the data structure of the DCT coefficient 105 of the 8 × 8 pixel block. As shown in the figure, one block is composed of a DC coefficient representing an average luminance level of 8 × 8 = 64 pixels and an AC coefficient representing a horizontal and vertical frequency distribution of an image.

Generally, moving images such as TV signals are temporally
The spatial correlation is strong, and when DCT is applied to transform the data into frequency domain data, the components are concentrated and distributed in a relatively low frequency domain. On the other hand, the discrimination characteristic of the human eye is low frequency, that is,
Sensitive to flat images and insensitive to high frequencies, ie high definition images. Therefore, by finely quantizing the low frequency region and roughly quantizing the high frequency region, the quantization distortion can be concentrated in the high frequency region of the image, and visual image quality deterioration can be suppressed.

From the above, the image definition determination device 1
07 is the definition of the image data (Act
(ivity) is obtained and definition information corresponding thereto is provided to the quantization controller 109.

In the quantizer 108, the DCT coefficient 105
Under the control of the quantization control device 109, for example, as shown in FIG. 8, is divided into four areas from a low frequency area to a high frequency area, and the quantization steps are performed in the order of area 0 <area 1 <area 2 <area 3. Coarse and quantize. The buffer device 106 delays the data by a time corresponding to the delay until the quantization level is controlled based on the DCT-converted data of the DCT coefficient 105.

The encoder 110 zigzag-scans the two-dimensionally quantized data obtained from the quantizer 108 from the lower spatial frequency to the higher spatial frequency to make it one-dimensional, and then sets the zero coefficient. In the run-length coding, the non-zero coefficient is variable-length coded in the two-dimensional Huffman coding, and the coded data 112 is output. In run-length coding, lossless compression is performed by counting zero runs, and in Huffman coding, short code words are assigned to data with a high probability of occurrence,
By allocating long codewords to data with a low probability of occurrence, the amount of information is reduced on average.

[0009]

As described above, in the conventional image coding apparatus, the quantization control is performed based on the definition information of the image data. However,
When quantizing a color signal, the I-axis (orange-cyan axis) system and Q
Since there is a difference in human visual characteristics from the axis (yellow green-purple axis) system, it is not a good idea to treat them both equally. In particular, there is a problem that deterioration of image quality is easily noticeable in an image having a reddish hue that is easily perceived by humans.

The present invention has been made to solve the above problems, and it is possible to reduce the deterioration of the image quality due to the hue and further improve the coding efficiency and the image coding apparatus and the image coding. The purpose is to obtain a method of conversion.

[0011]

In order to solve the above-mentioned problems, an image coding apparatus of the present invention is an orthogonal transform that obtains transform coefficients by orthogonally transforming image data input for each block consisting of a plurality of pixels. Means, quantizing means for quantizing the conversion coefficient, definition determining means for determining the definition of the image based on the conversion coefficient, and color determination for determining whether the image data indicates a specific color. Means, quantization control means for controlling the quantization characteristic of the quantization means according to the determination result of the color determination means and the determination result of the definition determination means, and data quantized by the quantization means. An image encoding device having:
The quantization control means may perform coarse quantization on the image data determined to have a low definition by the definition determination means regardless of the determination result of the color determination means.

In order to solve the above-mentioned problems, the image coding method of the present invention comprises an orthogonal transforming step of transforming image data input for each block consisting of a plurality of pixels to obtain transform coefficients, and the transform coefficient. A fineness determination step of determining the fineness of the image based on the following, a color determination step of determining whether the image data indicates a specific color, a determination result of the color determination step, and a determination of the fineness determination step An image encoding method having a quantization step of controlling a quantization characteristic according to a result, a quantization step of quantizing the transform coefficient, and an encoding step of encoding the data quantized in the quantization step, In the quantization step, coarse quantization is performed on the image data determined to have a low definition in the definition determination step, regardless of the determination result of the color determination step.

[0013]

[0014]

[0015]

1 is a block diagram best representing the features of the present invention. The portions corresponding to those in FIG. 6 are designated by the same reference numerals and the description thereof will be omitted. In FIG. 1, reference numeral 201 denotes a color determination device, which performs color determination on the image data 102 from the memory device 101 based on the definition information of the definition determination device 107. Next, the operation of this embodiment will be described. Image data 100
Of the image data 102 for each block supplied from the memory device 101, the color difference data BY / RY is referred to as CB / CR. FIG. 2 shows an embodiment of the color determination device 201, which includes a matrix conversion device 300 and a comparator 3.
03, 304, a counter 306, etc.

In FIG. 2, the blocked color difference image data CB and CR are θ by the matrix conversion device 300.
Receive the coordinate transformation of °. The coordinate axes after conversion are CB ', C
R '

[0017]

[Equation 1]

It can be expressed as For example, Q = 45
In the case of coordinate conversion of °, CB '= (√2 / 2) ・ (CB + CR) CR' = (√2 / 2) ・ (-CB + CR) ... (2), which can be realized by simple addition and subtraction. .

The color difference data 301 and 302 of CR 'and CB' after coordinate conversion are input to the comparators 303 and 304 to be compared with the preset threshold value. The state of this area comparison is shown in FIG. In the example of FIG. 3, from the color difference data 301 and 302, Red (104
°) and Magenta (61 °) color signals are determined by providing threshold values a and b on the coordinate axes CR ′ and CB ′ converted at Q = 45 ° for area comparison. Here, when the threshold value a for the CR ′ axis and the threshold value b for the CB ′ axis are set to CR ′> a, CB ′> b, the color difference data corresponding to the shaded area including Red and Magenta in FIG. Can be extracted.

The hue area designation can be realized by coordinate conversion at an arbitrary angle and at least one threshold value comparison.

FIG. 4 shows the relation of i <with respect to the CR ′ axis and the CB ′ axis whose coordinates are rotated by θ in order to detect Red and Magenta.
The case where one or more threshold values such as CR '<j, k <CB' are set is shown.

In this way, after the pixel data having the corresponding color in the area is extracted, the counter device 306 integrates it. Here, for example, as shown in FIG. 5, as the threshold value 307 (FIG. 2) of at least one or more, TH1,
TH2 is provided, and the number of pixels corresponding to the number of pixels in a block (8 × 8 = 64 pixels) is counted. Figure 5
When the number of pixels corresponding to the shaded area (1 ≦ TH1 <the number of pixels in the block <TH2 ≦ 64) is detected, the color determination result of the block is activated.

Next, color determination is performed by combining the color determination result and the image definition information 203 obtained from the definition determination device 107 of FIG. That is, even if the color of the image block has a specific color, if a low-frequency image (flat image) is determined from the frequency distribution, max | AC | 203 in FIG.
Is the Low level, and the color determination information 202 at this time is L
The ow level is output. When the definition information 203 determines a high-frequency image, max | AC | 203 is Hi.
The gh level is taken, and the color determination information 202 is output as the High level.

Therefore, in the quantization controller 204, when the color difference image data is quantized, the color determination information 202 is set to High.
When h, the quantization step is controlled so that finer quantization levels are assigned to the surrounding DCT blocks. On the contrary, when the color determination information 202 is Low, the quantization step control does not depend on the color determination information 202.

As described above, according to this embodiment, in order to determine the color of the image block data, the color difference data BY / Y /
The RY coordinate conversion is performed, the area of a specific hue is determined by a comparator, the color information is detected, and the DCT is detected even with the specific color data based on the detection result and the definition information.
When max | AC | of the subsequent AC coefficient is low and the image is flat, the quantization step is controlled so that an unnecessary code amount is not allocated, thereby improving coding efficiency. You can In addition, the color determination device 20
With No. 1, it is possible to perform arbitrary hue determination without significantly increasing the amount of hardware.

[0026]

As described above, according to the present invention,
For image data that is determined to have low definition, quantization control that does not depend on the determination result of color determination is performed and quantization is performed, so even if block image data of a specific color is used, a flat image is obtained. In that case, by controlling the quantization step so as not to allocate more code than necessary, there is an effect that the coding efficiency can be improved.

[0027]

[0028]

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a block diagram showing an embodiment of the color determination device of FIG.

FIG. 3 is an explanatory diagram of a hue determination area.

FIG. 4 is another explanatory diagram of a hue determination area.

FIG. 5 is an explanatory diagram for performing pixel integration within a block.

FIG. 6 is a block diagram showing a conventional image encoding device.

FIG. 7 is a schematic diagram showing a DCT coefficient distribution.

FIG. 8 is an explanatory diagram for performing quantization area division.

[Explanation of symbols]

102 image data 103 DCT device 104 AC coefficient 105 DCT coefficient 107 Definition determination device 108 Quantizer 110 encoder 201 Color judgment device 202 Color judgment information 203 Definition information 204 Quantization control device

Continuation of front page (56) Reference JP-A-7-288845 (JP, A) JP-A-7-288809 (JP, A) JP-A-6-276476 (JP, A) JP-A-6-225340 (JP , A) JP 6-153234 (JP, A) JP 5-284487 (JP, A) JP 5-236511 (JP, A) JP 5-227441 (JP, A) JP 5-153405 (JP, A) JP 5-95536 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H04N 11/00-11/22 H04N 7/24-7 / 68

Claims (5)

(57) [Claims] 1. An orthogonal transform means for orthogonally transforming image data input for each block composed of a plurality of pixels to obtain transform coefficients, a quantizing means for quantizing the transform coefficients, and based on the transform coefficients. A definition determination unit that determines the definition of an image, a color determination unit that determines whether the image data indicates a specific color, a determination result of the color determination unit and a determination result of the definition determination unit And a quantization control means for controlling the quantization characteristic of the quantization means, and an encoding means for encoding the data quantized by the quantization means, the image encoding device comprising: The image formation control means applies the color to the image data determined to have a low definition by the definition determination means.
An image coding apparatus, characterized in that rough quantization is performed regardless of the determination result of the determination means . 2. The image coding apparatus according to claim 1, wherein the definition determining unit determines the definition of an image from a distribution state of AC coefficients included in the conversion coefficient. 3. The color determination means includes a calculation means for performing a matrix calculation of color difference data obtained from the image data, and a comparison means for comparing at least one threshold value with a calculation result of the calculation means. The image coding apparatus according to claim 1 or 2, further comprising: a counter unit that integrates the comparison results of the comparison unit and compares the integration result with at least one threshold value. . 4. The image coding apparatus according to claim 3, wherein the color determination unit has a unit that specifies a region for an arbitrary hue by the calculation unit. 5. An orthogonal transforming step for obtaining transform coefficients by orthogonally transforming image data input for each block composed of a plurality of pixels, and a fineness determining step for determining fineness of an image based on the transform coefficients. , A color determination step of determining whether the image data indicates a specific color, controlling the quantization characteristic according to the determination result of the color determination step and the determination result of the definition determination step, the conversion coefficient A quantization step of quantizing, and an image coding method having a coding step of coding the data quantized in the quantization step, wherein in the quantization step, the definition in the definition determination step is For image data determined to be low ,
An image coding method characterized in that coarse quantization is performed regardless of the determination result of the color determination step .