JP2002238060A - Image-coding method, image coder, program and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image coder that enhances a reproduced image of an area of a specific color by quantizing the block of the specific color with a smaller quantization scale and to provide an image coding method, a program for the method and a recording medium for recording the program. SOLUTION: The image coder, that codes each block resulting from dividing an original image by a prescribed pixel number, is provided with a color discriminating means 34 that decides whether or not a color of each block is a prescribed color, a quantization scale revision means 18 that changes the quantization scale of each block, in response to an output of the color decision means 34 so that the quantization scale of the block in a prescribed color is smaller than the quantization scale of the block, whose color is not the prescribed color, and a quantization means 15 that quantizes the image of each block by each quantization scale decided by the quantization scale revision means 18.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image encoding apparatus and an image encoding method for improving the reproduction quality of a subject having a specific color by adaptively changing a quantization scale for a specific area having the specific color. About. Further, the present invention relates to a program for realizing the method and a recording medium on which the program is recorded.

[0002]

2. Description of the Related Art Technical standards for encoding image signals include, for example, JPEG and H.264. 26X series and MPEG.

[0003] JPEG is based on discrete cosine transform (hereinafter referred to as "DC
T ”. ), Quantization, and variable-length coding are standards for coding color still images.

[0004] H. The 26X series is a video coding standard for audio-visual communication applications such as video telephony and video conferencing, and uses DCT, quantization and variable-length coding in the forward direction to reduce redundancy between frames. Further combine motion compensated inter-frame prediction.

[0005] MPEG combines DCT, quantization, variable length coding, and forward motion compensated interframe prediction with bidirectional motion compensated interframe prediction, and enables random access. Several sheets (for example, 15
Group), a GOP (Group Of Pictures) structure is adopted. GOP
Are I-picture (intra-coded picture), P-picture (p
redictive-coded picture) and B picture (bidirect
ionally predictive-coded picture). I
A picture is a screen that is encoded from only that information and that is generated without using inter-frame prediction. A P picture is a screen generated by performing prediction from an I or P picture. A B picture is a screen generated by bidirectional prediction.

[0006] MPEG is MPEG-1, which standardizes encoding technology for storage media such as CD-ROM, broadcast, communication,
Functions such as MPEG-2 that standardizes the technology for encoding general-purpose digital video signals and accompanying audio signals regardless of the target media such as storage, content stream management, scene synthesis, and content-based copyright protection It has been standardized step by step to MPEG-4, which standardizes the encoding technology of enhanced bitrate and ultra-low bit rate. Above all, MPEG-2 is widely used for broadcasting and AV equipment.

[0007] An image coding apparatus for coding an image signal includes:
An orthogonal transformation circuit for transforming an image divided into blocks each having a predetermined number of pixels into data suitable for encoding by two-dimensional orthogonal transformation for each block, and a quantum circuit for quantizing the converted data on an appropriate quantization scale Circuit, a variable-length encoding circuit that performs variable-length encoding on the output of the quantization circuit according to a predetermined code table, and controls the quantization scale to an appropriate value in consideration of the output bit rate of the image encoding device. And a rate control circuit.

[0008] In particular, an image coding apparatus for MPEG-2 includes a screen rearrangement section, a subtraction section, a DCT section, a quantization section, a variable length coding section, a buffer memory, a rate control section, an inverse quantization section, and an inverse quantization section. DC
It is configured to include electronic circuits of a T unit, an addition unit, a frame memory, and a motion compensation unit.

An input video signal supplied from a video camera is divided into blocks of a predetermined number of pixels, and after being converted from an analog signal to a digital signal, the order of pictures is rearranged in an order suitable for encoding processing.

[0010] The output of the screen rearranging section is supplied to the DCT via the subtracting section.
DCT encoded in the unit, quantized at a predetermined bit rate in the quantization unit, variable-length code in the variable-length encoding unit, for example, encoded with a Huffman code, and
Output to the buffer memory. The buffer memory outputs encoded data at a predetermined bit rate.

In the case of I-pictures and P-pictures, the output of the quantization unit is also input to the inverse quantization unit, and is subjected to inverse DCT after inverse quantization since it is used as a reference screen in the motion compensation unit. An inverse DCT is performed in the section. Inverse DC
The output of the T section is added to the output of the motion compensation section by the addition section,
The data is input to the frame memory and then input to the motion compensator to be sequentially processed. The motion compensation unit performs forward prediction,
It performs backward prediction and bidirectional prediction and outputs the result to the addition unit and the subtraction unit.

The subtraction unit performs a subtraction between the output of the image rearrangement unit and the output of the motion compensation prediction unit, and performs a subtraction between the decoded video signal and the video signal decoded by the inverse quantization unit and the inverse DCT unit. Is calculated. Intra-frame coding (I picture)
In the case of, the subtraction unit does not perform the subtraction processing, and the data simply passes.

On the other hand, since the code amount generated in the variable length coding unit is variable, the rate control unit monitors the buffer memory to maintain a predetermined bit rate.
Controls the quantization operation of the quantization unit.

[0014] The rate control unit is configured to output a target bit amount (GOP) of the GOP.
A target bit amount is set, and a target bit amount (picture target bit amount) is assigned to each picture unit in the GOP target bit amount, and encoding is performed with a predetermined target bit amount. Bit allocation in a picture is performed by setting a quantization scale for each macroblock (hereinafter abbreviated as “MB”).

That is, generally, the above-described processing is performed in the following three steps.

In the first step, the amount of bits allocated to each picture in the GOP is allocated based on the amount of bits allocated to pictures that have not been encoded in the GOP, including the picture to be allocated. This distribution
It repeats in the order of the coded pictures in the GOP, and sets a picture target bit amount for each picture.

In the second step, a reference value of the quantization scale is set for each MB. That is, in the second step, the first
In order to match the bit amount allocated to each picture obtained in the step with the actual bit amount generated, the reference value of the quantization scale is set based on the capacity of three types of virtual buffers independently set for each picture type. Determined by feedback control in MB units.

In the third step, the quantization scale value is corrected based on the MB activity in MB units so as to reflect the visual characteristics. While maintaining the frame target bit amount, the quantization scale is corrected to be smaller than the reference value for MBs with low activity, and the quantization scale is corrected to be larger than the reference value for MBs with high activity. In this manner, adaptive quantization is performed in consideration of visual characteristics, particularly activity.

Such an MPEG is disclosed, for example, in "Synthetic Multimedia Selection MPEG" (edited by the Institute of Image Information and Television Engineers, Ohmsha).

It is generally known that a person has the following visual characteristics (1) to (3).

(1) The visual sensitivity of a high-frequency component is lower than that of a low-frequency component. (2) The gradation distortion is more conspicuous in gray parts than in white and black. (3) The perceived resolution is reduced for a sudden change in the image.

[0022]

In the above-described processing method in the third step, the quantization scale is changed using only the activity. Therefore, in an MB having a high activity, the quantization is always coarse regardless of the image content of the MB. And the reproduced image is degraded. Therefore, there is a problem that even if a part of the image that the photographer or the viewer is interested in is high and important, if the activity is high, quantization is coarsely performed.

The portions having high activity are, for example, the eyes and nose of the face of a person who is generally interested, such as a photographer, and the edge and periphery of an object. Further, since the difference value is consequently enlarged, the activity is also high in a portion where the luminance value is high (bright).

Accordingly, the present invention provides an image encoding apparatus, an image encoding method, and an image encoding method capable of designating a portion in an image to be finely quantized, that is, a portion in which the definition of a reproduced image is desired to be increased. It is an object of the present invention to provide a program to be realized and a computer-readable recording medium on which the program is recorded.

[0025]

According to a first aspect of the present invention, in an image encoding apparatus for encoding an original image for each block obtained by dividing the original image by a predetermined number of pixels, each block has a predetermined color. Color determination means for determining whether or not each of the blocks has a predetermined color in accordance with an output of the color determination means such that a quantization scale in a block of a predetermined color is smaller than a quantization scale in a block which is not a predetermined color. It is provided with quantization scale changing means for changing the quantization scale, and quantization means for quantizing the image of each block at each quantization scale determined by the quantization scale changing means.

In the first means, the quantization scale changing means further changes the quantization scale of each block so as to substantially maintain the amount of encoded data allocated to the original image. May be.

In the first means, the color determining means determines a predetermined color for each pixel for each block and determines a ratio of a pixel occupied by the predetermined color to the total number of pixels in the block. In addition, the quantization scale changing means may further change the quantization scale of a block of a predetermined color in accordance with the ratio such that the quantization scale decreases as the ratio increases.

Then, in such a first means,
The color determination means may determine the color of the block using an average value of the color difference signals for all pixels in the block.

Further, in such a first means,
The quantization scale changing unit may further change a quantization scale for a peripheral block of a block determined to be a predetermined color to a quantization scale smaller than a quantization scale of a block that is not a predetermined color.

In the first means, the quantization scale changing means may change the quantization scale when the quantization scale is changed and when the block is within a predetermined area on the screen. May be.

Then, in such a first means,
The predetermined color may be set as a skin color.

Further, in such a first means,
The image processing apparatus may further include activity calculation means for obtaining an image activity, wherein the quantization scale changing means may change the quantization scale according to the activity from the activity calculation means when changing the quantization scale.

The first means may further include a luminance average value calculating means for calculating a luminance average value of each block, and the quantization scale changing means may further include a step for changing the quantization scale. The quantization scale may be changed according to the average luminance value from the average luminance value calculation means.

Then, in such a first means,
The image processing apparatus further includes a residual value calculating unit that obtains a residual value of a motion vector in the predicted image data, wherein the quantization scale changing unit further includes a residual value from the residual value calculating unit when the quantization scale is changed. The quantization scale may be changed according to.

According to the second means of the present invention, an image divided into blocks each having a predetermined number of pixels is divided into two blocks for each block.
Orthogonal transformation means for transforming into data suitable for encoding by dimensional orthogonal transformation, color determination means for determining whether the color of each block is a predetermined color, and a quantization scale for the block of the predetermined color having a predetermined value. The quantization scale changing unit that changes the quantization scale of each block according to the output of the color determination unit so as to be smaller than the quantization scale of the block that is not a color, and the quantization scale changing unit determines the quantization scale. An image encoding apparatus comprising: a quantization unit that quantizes an image of each block at each quantization scale; and a variable-length encoding unit that performs variable-length encoding on an output of the quantization unit according to a predetermined code table. Is configured.

According to a third aspect of the present invention, in an image encoding method for encoding an original image for each block obtained by dividing the original image by a predetermined number of pixels, it is determined whether or not each block has a predetermined color. In the color determination step, the quantization scale of each block is changed according to the determination result of the color determination step so that the quantization scale in the block of the predetermined color is smaller than the quantization scale in the block that is not the predetermined color. And a quantization step of quantizing an image of each block with each quantization scale determined in the quantization scale changing step.

According to a fourth aspect of the present invention, in an image encoding program for causing a computer to encode an original image for each block divided by a predetermined number of pixels, it is determined whether or not each block has a predetermined color. A color determination step for determining;
A quantization scale change that changes the quantization scale of each block according to the result of the color determination step so that the quantization scale of the block of the predetermined color is smaller than the quantization scale of the block that is not the predetermined color. And a quantization step of quantizing an image of each block at each quantization scale determined in the quantization scale changing step.

According to a fifth aspect of the present invention, in a computer-readable recording medium recording an image encoding program for causing a computer to encode an original image in blocks divided by a predetermined number of pixels, the color of each block is A color determination step of determining whether or not the color is a predetermined color; and a determination result of the color determination step such that a quantization scale of the block of the predetermined color is smaller than a quantization scale of the block which is not the predetermined color. A quantization scale changing step of changing the quantization scale of each block according to the following, and a quantization step of quantizing an image of each block at each quantization scale determined in the quantization scale changing step. Is done.

According to a sixth aspect of the present invention, there is provided a photographing means for photographing an image of a subject, converting the image into a digital video signal, and outputting the digital image signal; an image encoding means for compressing and encoding image data from the photographing means; Recording means for recording the encoded data encoded by the image encoding means on a recording medium, wherein the image encoding means comprises:
Means.

These image encoding apparatuses, image encoding methods,
The image coding program, the recording medium storing the program, and the digital camera can determine a specific color area on the screen, and quantize the specific color area finer than other color areas. Quantize on scale. For this reason, by setting the areas of high interest or important parts of the photographer or viewer with colors, such parts can be finely quantized. Can be improved.

[0041]

Embodiments of the present invention will be described below with reference to the drawings. In each of the drawings, the description of the same configuration may be omitted.

(First Embodiment) FIG. 1 is a block diagram showing the configuration of an image coding apparatus according to the first embodiment.

In FIG. 1, an image coding apparatus according to the first embodiment includes an image rearrangement unit 11, a subtraction unit 12, a switch 13, a DCT unit 14, a quantization unit 15, and a variable length coding unit 1.
6, buffer memory 17, control unit 18, inverse quantization unit 2
1, the inverse DCT unit 22, the adding unit 23, the frame memory 24,
26, a motion compensator 25, a motion detector 27, and a skin color component detector 34.

The input video signal is input to the image rearranging section 11, and after converting the analog signal into a digital signal, the order of the pictures is rearranged in an order suitable for the encoding process. That is, the image rearranging unit 11 encodes the I picture and the P picture first, and then rearranges the B picture in an order suitable for encoding.

The image rearranging section 11 divides the image into blocks each having a predetermined number of pixels, for example, 16 × 16 MBs, and outputs each MB. This output is input to the DCT unit 14 via the subtractor 12 and the switch 13, and DCT coding is performed. DCT
Is one of the orthogonal transformations, and the other is Hadama
rd) transformation, Karhunen-Loeve transformation, Fourier transformation and the like.

The output of the DCT unit is input to the quantization unit 15 and is quantized on the quantization scale (quantization step) specified by the control unit 18. The quantization scale is a parameter for controlling the amount of generated bits by scaling the quantization characteristics. The quantization characteristic is displayed as an 8 × 8 matrix in order to set the relative quantization accuracy between DCT coefficient values in the block. For example, an intra macroblock quantization matrix or a non-intra macroblock quantization matrix Are prepared. In the present embodiment, as will be described later, the control unit 18 uses the correction coefficient Ncolor in consideration of the MB characteristic for the quantization scale reference value refQ (j).
(j) is multiplied to calculate the quantization scale Q (j). The control unit 18 is a circuit that controls the output bit rate, and also notifies the inverse quantization unit 21 of the specified quantization scale.

The output of the quantization unit 15 is the variable length coding unit 1
6 and the inverse quantization unit 21. The variable-length encoding unit 16 encodes a variable-length code, for example, a Huffman code that assigns a shorter code to data with a higher appearance frequency, and outputs the encoded data to a buffer memory 17 of a memory. The buffer memory 17 outputs the encoded data at a predetermined bit rate as an output of the image encoding device. The control unit 18 controls the quantization operation of the quantization unit 15 so as to maintain a predetermined bit rate by monitoring the buffer memory 17 because the code amount generated in the variable length coding unit 16 is variable. .

On the other hand, when the output of the quantization unit 15 is an I picture and a P picture, the output of the quantization unit 15 is also input to the inverse quantization unit 21 because the output is used as a reference screen by the motion compensation unit 25. Is done. The input signal is transmitted to the control unit 18
Inverse after being quantized with the quantization scale notified by
The signal is input to the DCT unit 22 and inverse DCT is performed. Reverse DCT section 22
Is added to the output of the motion compensator 25 by the adder 23, and is input to the frame memory 24. Then, the output of the screen rearranging unit 11 is also input to the frame memory 26 and the motion detecting unit 27.

The motion detecting section 27 uses the image data of the time to be compared stored in the frame memory 26 and the image data of the current time from the screen rearranging section 11 to determine the motion vector of each MB (hereinafter referred to as “MV”). Is abbreviated as "). For example, the motion detection unit 27 performs block matching between a current frame and a predicted frame for each MB within a predetermined search range for each MB, and detects a motion amount that minimizes a prediction error. Detect MV.

As shown by the broken lines in FIG. 1, the decoded image data of the time to be compared and stored in the frame memory 24 and the image of the current time from the screen rearranging unit 11 are stored without using the frame memory 26. From the data, the MV of each MB
May be calculated respectively.

The MV calculated by the motion detecting section 27 is input to the motion compensating section 25 and reconstructs a predicted image using the MV based on the decoded image data of the time to be compared stored in the frame memory 24. I do. Then, the motion compensation unit 25
Outputs a predicted image to the addition unit 23 and the subtraction unit 12.

The inverse quantization section 21, inverse DCT section 22, addition section 23, frame memories 24 and 26, and motion compensation section 25
And the motion detection unit 27 constitutes a local decoding unit.

The subtraction unit 12 performs a subtraction between the output of the screen rearranging unit 11 (image data of the current time) and the output of the motion compensation unit 25 (image data of the time to be compared) to form a prediction error. I do. In the case of intra-frame coding (I picture), the switch 13 allows the subtractor 12 to simply pass data without performing the subtraction processing.

That is, the switch 13 is connected to the screen rearranging unit 1.
When the output of 1 is an I picture, the terminal b and the terminal c of the switch 13-a are connected, and the terminal f and the terminal g of the switch 13-b are connected. When the output of the screen rearranging unit 11 is a P picture and a B picture, the switch 13
The terminal a and the terminal c of the switch 13-a are connected, and the terminal e and the terminal g of the switch 13-b are connected.

Further, the output of the screen rearranging section 11 is also input to the skin color component detecting section 34.

The skin color component detecting section 34 determines, for each MB, whether or not the MB is a skin color based on the color difference signals Cb and Cr. The skin color component detection unit 34 outputs the result of the determination to the control unit 18.
Output to The control unit 18 sets a target bit amount for each picture, and sets a quantization scale reference value for each MB. Then, the control unit 18 corrects the quantization scale based on the output of the skin color component detection unit 34.

Next, main processing of the image encoding apparatus will be described. FIG. 2 is a flowchart illustrating main processing of the image encoding device according to the first embodiment.

In FIG. 2, the control unit 18 takes in the output h (j) of the skin color component detection unit 34 (St1).

For example, when the color difference signals Cr and Cb of a pixel satisfy Expression 1, the skin color component detection unit 34 determines that the pixel is a skin color. FIG. 3A shows a region to be a flesh color. And
The determination of Expression 1 is performed for all pixels belonging to the j-th MB, and the total number of pixels determined to be flesh color is output to the control unit 18 as h (j). k1 × Cb ≦ Cr ≦ k2 × Cb (1) Here, k1 and k2 are constants. In the case of flesh color, for example, k1 and k2 are −0.69 ≦ k1 ≦ −0.45 − 2.69 ≦ k2 ≦ −1 (2) Any value in the range. As a result of the experiment, race (white race, yellow race, black race, and their mixed race)
It is preferable that k1 and k2 be set to k1 = −7 / 16 and k2 = −1 in order to determine the skin color of a person well regardless of the difference.

Further, by appropriately selecting these constants, it is possible to arbitrarily set an object of high interest such as a photographer or a viewer. For example, when setting to whales or sea pigs, k1 and k2 may be selected so that the body color is gray. In addition, for example, in the case of setting to a swan, a crane, or the like, k1 and k2 may be set so that the body color is white.

An external input unit is further provided so that
By designating 1, k2 from the external input unit, k1, k2 may be set. As a result, the set value can be appropriately changed, so that when the photographer or the like wants to change the object of interest, the change can be made in accordance with the change.

In such a case, for example, as shown in FIG. 3B, k3 ≦ Cb ≦ k4 k5 ≦ Cr ≦ k6 (3) Here, k3 to k6 are constants. In addition,
Although the predetermined color region is exemplified in Expressions 1 and 3, the expression representing the color region depends on the shape of the line indicating the boundary of the designated color.

Returning to FIG. 2, next, the control unit 18 generates an evaluation value H (j) of the MB (St2). For example, the control unit 18
Determines the evaluation value H (j) indicating whether or not the MB is a flesh-colored area according to a rule according to Equation 4. If h (j)> th1_h, H (j) = a1 if th1_h ≧ h (j)> th2_h, then H (j) = a2 th2_h ≧ h (j), then H (j) = a3 … (4)

Next, the control unit 18 determines whether St1 and St2
Is performed for all MBs in the picture (St
3). Next, the control unit 18 reads the generated bit amount (St4). Next, the control unit 18 sets a target bit amount (picture target bit amount) for each picture, and sets a quantization scale reference value refQ (j) for each MB (St).
5).

Next, the control unit 18 sets the correction coefficient Ncolor (j)
Is generated according to Equation 5 (St6). Ncolor (j) = (R × H (j) + avg_H) / (H (j) + R × avg_H) (5) Next, the control unit 18 calculates the obtained correction coefficient Ncolor (j) into Expression 6
To correct the quantization scale Q (j) for the MB of the subject (St7). Q (j) = refQ (j) × Ncolor (j) (6) Next, the control unit 18 executes St6 and St7 for all MBs in the picture (St8).

As described above, the image coding apparatus according to this embodiment discriminates the subject in the entire image by detecting the MB including the characteristic color of the subject of interest to the photographer or the viewer. I do. Then, the image encoding apparatus uses the correction coefficient to allocate the quantization scale value for the detected MB to a value smaller than the other MBs, thereby substantially maintaining the picture target bit amount allocated to the picture. While improving the reproduction image quality of a specific subject.

Here, the information processing by the equations (4) to (6) is processing corresponding to the third step described in the related art. th1_h and th2_h are threshold values for determining the magnitude of the correction coefficient according to the number of skin color pixels existing in the MB and satisfying Expression 1, and the present embodiment corrects according to the number of skin color pixels existing in the MB. The coefficient is divided into three stages. a1, a2, a3 are
As can be seen from equations (4) and (5), it is a value related to the magnitude of the correction coefficient, and a1 is a value for the MB determined to be flesh color.
(j) is set to be small, and a3 is a value for the MB not determined to be a flesh color, so that the quantization scale value Q (j) obtained by Expression 6 is set to be large. a2
Indicates that part of the MB, such as the outline of the subject, was judged to be skin tone
This is a value for MB and is an intermediate value between a1 and a3.

For example, assuming that th1_h = 30 and th2_h = 10,
a1: a2: a3 = 1: 10: 100. Alternatively, for example, it is assumed that th1_h = 50 and th2_h = 30, and a1: a2: a3 = 1:
5: 200. In the present embodiment, the data is divided into three stages, but may be divided into two stages, four stages, or the like, and can be divided into an arbitrary number of stages.

Avg_H is the average value of H (j) in the picture, and R is the correction coefficient Ncolor (j) as shown in equation (7).
Determine the range of

1 / R ≦ Ncolor (j) ≦ R (7) For example, if R = 2, the correction coefficient Ncolor is 0.5 to
2 range. Therefore, when the range of the case where the definition is corrected to be high and the case where the correction is lowered are widened, the value of R may be increased.

RefQ (j) is a reference quantization scale in the j-th MB, and is obtained by Expressions 8 and 9. The information processing for obtaining refQ (j) corresponds to the second step described in the related art. In the case of an I picture, refQ (j) = di (j) × 31 / r (8-1) di (j) = d0i + B (j-1) −Ti × (j-1) / MBnum (9- 1) In the case of a P picture, refQ (j) = dp (j) × 31 / r (8-2) dp (j) = d0p + B (j−1) −Tp × (j−1) / MBnum (( 9-2) In the case of a B picture, refQ (j) = db (j) × 31 / r (8-2) db (j) = d0b + B (j−1) −Tb × (j−1) / MBnum ... (9-3) di (j), dp (j), and db (j) are the virtual buffer occupancy of each picture obtained by Expression 9, and d0i (j), d0p (j), d0b (j) Is
This is the initial value of the virtual buffer capacity for each picture.
B (j) is the amount of generated bits from the beginning of the picture to the j-th MB. Ti, Tp, and Tb are picture target bit amounts for each picture obtained in the first step described in the related art. In B (j), MBnum is the total number of MBs in one picture. r is a reaction parameter for controlling the response speed of the feedback loop, and is a value obtained by dividing the bit rate by the picture rate and doubling it.

The quantization scale of MB is given by Equations 5, 6, and 8.
And Equation 9, even if the quantization scale for a specific color MB is assigned smaller than the quantization scale for a color MB excluding a specific color, Equations 5, 6,
The picture target bit amount is maintained within the error (within several percent) that can occur in Equations 8 and 9. Therefore, the reproduction image quality of a specific subject is improved while substantially maintaining the picture target bit amount allocated to the picture.

In this embodiment, H (j) of MB can be calculated over the entire screen before encoding.
For this reason, the average avg_H in the screen can use not the avg_H of the immediately preceding picture but the avg_H of the picture. Therefore, it is possible to follow a specific subject even with a rapidly moving image.

In the above-described embodiment, the flesh-color component detection unit 34 determines whether or not the skin color is a pixel-by-pixel unit according to Equation 1. Then, the flesh-color component detection unit 34 evaluated the percentage of the MB in the flesh-colored area using a1, a2, and a3 according to Equation 4 according to the result. On the other hand, in this processing, the image encoding device calculates
Expression 10 may be used instead of, and a flesh color component detection unit 34 that uses Expression 11 instead of Expression 4 may be provided.

K1 × [Cb] ≦ [Cr] ≦ k2 × [Cb] (11) Here, [Cb] is the average value of Cb for all pixels in the MB. [Cr] is the average value of Cb for all pixels in the MB. If MB satisfies Equation 11, H (j) = a1 If MB does not satisfy Equation 11, H (j) = a3 (12)

As described above, in order to determine whether or not a pixel is in a flesh color area in MB units, not in pixel units, the flesh color component detection unit 34 determines the intermediate value as in the case of H (j) = a2 in equation (4). Although no judgment can be made, the circuit configuration can be simplified and the processing time can be shortened.

In particular, for an MB including a contour portion of a subject,
Since the front and rear MBs are often either inside or outside the subject, the flesh color component detection unit 34 normally makes a determination using Equations 1 and 4, and when MB is H (j) = a2 If there is, the next MB is determined by Expressions 11 and 12. Then, the flesh-color component detection unit 34 sets the MB in which H (j) = a3.
Are determined again by the equations (1) and (4) when. At this time, the evaluation is performed again for consecutive MBs. In this way, the processing time can be reduced with the same image quality improvement effect as when only Expressions 1 and 4 are used.

Then, the processing of the skin color component detecting section 34 may be improved as follows. From the aforementioned human visual characteristics,
When the image quality deterioration in the peripheral portion of the subject of interest is large, the image quality deterioration is conspicuous. So, MB
Belongs to the skin color area, that is, H (j) = a1 or H
When (j) = a2, the skin color component detection unit 34 forcibly also corrects the eight peripheral MBs around the MB so that the reproduced image is improved as compared with the case of a3. ). In other words, if this case is a4, a4 is
a1 <a2 ≦ a4 <a3. For example, a4 is a1: a2: a4: a3
= 1: 10: 10: 100 or a1: a2: a4: a3 = 1: 1
0: 30: 100 is set.

By performing such processing, the image coding apparatus can reduce erroneous detection at the boundary of the subject or the like near the photographer or the like, or reduce the deterioration of the reproduced image. Visual quality can be improved.

Further, the processing of the skin color component detecting section 34 may be improved as follows. Generally, as a human psychology, a photographer often places a subject of interest in the center of a screen, and a viewer often focuses on the center of the screen. Therefore, the skin color component detection unit 34 determines whether
The processing of the correction coefficient is performed in consideration of the position of the MB.
For example, Equation 13 is used instead of Equation 4. When h (j)> th3_h and s1 ≦ MB_x ≦ s2 and s3 ≦ MB_y ≦ s4, H (j) = a5 When h (j)> th3_h, H (j) = a6 th3_h ≧ h (j) In the case of, H (j) = a7 (13)

Here, similarly to the above, th3_h is a threshold value for determining the magnitude of the correction coefficient in accordance with the number of pixels of the skin color that satisfies Expression 1 existing in the MB, and a5, a6 and a7 satisfy a5 ≦ a5. a6
<A7 is an arbitrary constant. When the position of the MB is represented by xy orthogonal coordinates, MB_x is the position of the MB in the x-axis direction, and MB_y is the position of the MB in the y-axis direction. s1, s2, s3 and s4 are arbitrary constants.

By performing the above processing, a specific color area is set in the reserved area (s1 ≦ MB_x ≦ s2 and s3 ≦ MB_y ≦ s4) reserved in the screen while substantially maintaining the picture target bit amount. If there is, the image quality of the color area can be efficiently improved. For example, by setting the reserved area at the center of the screen, when a large number of people are shown, only the main image quality can be efficiently improved.

Next, the image decoding apparatus will be described.

FIG. 4 is a block diagram showing an image decoding apparatus. The image decoding device includes a buffer memory 4 of a memory.
1, variable length code decoding section 42, inverse quantization section 43, inverse DCT section 44, addition section 45, frame memory 48, motion compensation section 4
9 and the electronic circuits of the screen rearranging section 46.

The encoded data (video elementary stream) is temporarily stored in the buffer memory 41 and input to the variable length decoding unit 42. The variable length decoding unit 42 decodes the macroblock encoded information, and separates the quantized information including the encoding mode, the MV, and the quantization scale, and the quantized DCT coefficients. The MV and the like are output to the motion compensation unit 49. The quantized DCT coefficients are restored to DCT coefficients by the inverse quantization unit 43 based on the quantization scale, and are converted to pixel space data by the inverse DCT unit 44. The addition unit 45 adds the output of the inverse quantization unit 44 and the output of the motion compensation unit 49, but does not add the I-picture when decoding it. All MBs in the screen are decoded, and the screen is rearranged in the original input order by the screen rearranging unit 46, converted into an analog signal, and output.
In addition, in the case of an I picture and a P picture, the output of the adder 45 is used as a reference screen in a subsequent decoding process, and is therefore stored in the frame memory 48 and output to the motion compensation unit 49.

The image decoding apparatus may process the output of the screen rearranging section 46 with a filter (FIL) 47 indicated by a broken line in order to smooth the reproduced image. FIL
47 smoothes the level difference when a reproduced image is mosaic due to a large difference in signal level between adjacent MBs.

Such an image encoding device and an image decoding device can be provided as an image compression / decompression device in an image device that captures an image and processes an image signal, such as a camera-integrated digital recording / reproducing device or a digital still camera. Further, the present invention installs the program in a computer via a recording medium in which the program for implementing one or both of the image compression process and the image decompression process is recorded, so that the computer can be used as an image encoding device, an image decoding device, and the like. It can also be a device or an image compression / decompression device.

Hereinafter, as an example, a camera-integrated digital recording / reproducing apparatus equipped with the image encoding apparatus according to the present invention will be described.

FIG. 5 is a block diagram showing the structure of a camera-integrated digital recording / reproducing apparatus.

In FIG. 5, a camera-integrated digital recording / reproducing apparatus includes a video encoder 111, an audio encoder 112, a video decoder 113, and an audio decoder 11.
4. File generator 115, file decoder 116, memories 117 and 120, memory controller 118, system control microcomputer 119, error correction code / decoder 12
1, drive control microcomputer 122, data modem 12
3, a magnetic field modulation driver 124, an operation unit 126, a servo circuit 130, a motor 131, a magnetic field head 132, and an optical pickup 133.

The video signal is supplied from a video input terminal to a video encoder 111, and is compressed and encoded. The audio signal is sent from the audio input terminal to the audio encoder 1
12 and compression-encoded. For example, an image captured by a video camera that generates a video signal by supplying imaging light of a subject to an imaging element such as a CCD (Charge Coupled Device) by an optical system (not shown) is input to the video input terminal. Supplied as Audio collected by the microphone is supplied to the audio input terminal as an audio signal.

As the video encoder 111, the image encoding device according to the present invention is used.

The audio encoder 112 is, for example, an MPEG
In the case of / Audio layer 1 / layer 2, each electronic circuit such as a subband encoding unit and an adaptive quantization bit allocating unit is provided. The audio signal is divided into 32 sub-band signals by the sub-band encoding unit, quantized by the adaptive quantization bit allocation unit according to the psychological auditory weighting, formed into a bit stream, and output. In order to improve coding quality, MPEG / Au
In the case of dio layer 3, the adaptive block length modified discrete cosine transform unit, the aliasing reduction butterfly unit,
A non-linear quantization unit and a variable length coding unit are introduced.

The output of the video encoder 111 and the output of the audio encoder 112 are supplied to a file generator 115. The file generator 115 has a file structure that can be handled by computer software that can play back video, audio, text, etc. synchronously without using a specific hardware configuration.
Convert the data structure of the video elementary stream and the audio elementary stream. Such software is, for example, QuickTime. Then, the file generator 115 multiplexes the encoded video data and the encoded audio data. The file generator 115 is provided by the system control microcomputer 1
19.

QuickT which is the output of the file generator 115
The ime movie file is sequentially written to the memory 117 via the memory controller 118. The memory controller 118 reads a QuickTime movie file from the memory 117 when the system control microcomputer 119 requests data writing to the recording medium 140.

Here, the transfer rate of the QuickTime movie encoding is set to a transfer rate lower than the transfer rate of the write data to the recording medium 140, for example, １ ／. Therefore, while the QuickTime movie file is continuously written to the memory 117, the reading of the QuickTime movie file from the memory 117 is performed in the memory 1
17 is monitored intermittently by the system control microcomputer 119 so as not to overflow or underflow.

QuickTime read from memory 117
The movie file is supplied from the memory controller 118 to the error correction code / decoder 121. The error correction encoder / decoder 121 writes the QuickTime movie file into the memory 120 once, and interleaves (in
terleaved) and redundant data of an error correction code. The error correction code / decoder 121 reads out the data to which the redundant data has been added from the memory 120 and supplies the data to the data modulator / demodulator 123.

When recording digital data on the recording medium 140, the data modulator / demodulator 123 modulates data so as to facilitate clock extraction during reproduction and to prevent problems such as intersymbol interference. For example, (1,7) RLL (ru
(n length limited) code or trellis code.

The output of the data modulator / demodulator 123 is supplied to the magnetic field modulation driver 124 and the optical pickup 133. The magnetic field modulation driver 124 applies a magnetic field to the recording medium 140 by driving the magnetic field head 132 according to the input signal. The optical pickup 133 irradiates the recording medium 140 with a recording laser beam according to an input signal. In this way, data is recorded on the recording medium 140.

The recording medium 140 is a rewritable optical disk, for example, a magneto-optical disk (MO, magneto-optical disk).
disk), phase change type disks, and the like. Recording medium 140
Is rotated by a motor 131 at a constant linear velocity (CLV), a constant angular velocity (CAV), or a zone CLV (ZCLV).

The drive control microcomputer 122 responds to a request from the system control microcomputer 119 by using the servo circuit 130.
Output the signal. The servo circuit 130 controls the entire drive by controlling the motor 131 and the optical pickup 133 according to the output. For example,
The servo circuit 130 performs radial servo, tracking servo, and focus servo of the recording medium 140 with respect to the optical pickup 133, and controls the rotation speed of the motor 131.

The system control microcomputer 119 includes:
An operation unit 126 for a user to input a predetermined instruction is connected. For example, in the case of an image coding apparatus capable of making k1 and k2 variable, these numerical values are input.

On the other hand, at the time of reproduction, the optical pickup 13
Reference numeral 3 irradiates the recording medium 140 with a laser beam at an output for reproduction, and receives the reflected light with a photodetector in the optical pickup 133 to obtain a reproduction signal. In this case, the drive control microcomputer 122 detects a tracking error and a focus error from the output signal of the photodetector in the optical pickup 133, and adjusts the read laser beam so as to be positioned on the track and focused on the track. The optical pickup 13 by the servo circuit 130
3 is controlled. Further, the drive control microcomputer 122
Controls the movement of the optical pickup in the radial direction in order to reproduce data at a desired position on the recording medium 140. The desired position is determined by the drive control microcomputer 122 by the system control microcomputer 119 in the same manner as during recording.
Are given and determined.

The reproduction signal of the optical pickup 133 is supplied to the data modulator / demodulator 123 and demodulated. The demodulated data is supplied to an error correction code / decoder 121,
The reproduced data is temporarily stored in the memory 120 and subjected to deinterleaving and error correction.
The QuickTime movie file after the error correction is stored in the memory 117 via the memory controller 118.

The QuickTime movie file stored in the memory 117 is output to the file decoder 116 in response to a request from the system control microcomputer 119. The system control microcomputer 119 reads the reproduction signal of the recording medium 140 from the memory 117 and reads the reproduction signal from the memory 117 and supplies it to the file decoder 116 in order to continuously reproduce the video signal and the audio signal. By monitoring the data amount, the memory controller 118 and the drive control microcomputer 122 are controlled so that the memory 117 does not overflow or underflow. Thus, the system control microcomputer 119 reads data from the recording medium 140 intermittently.

The file decoder 116 separates a QuickTime movie file into a video elementary stream and an audio elementary file under the control of the system control microcomputer 119. The video elementary stream is supplied to the video decoder 113, where it is subjected to compression encoding decoding, and is output as a video output from a video output terminal. The audio elementary stream is supplied to the audio decoder 114, where the audio elementary stream is subjected to compression encoding and decoding, and is output as an audio output from an audio output terminal. Here, the file decoder 116 outputs the video elementary stream and the audio elementary stream so that they are synchronized.

The video decoder 113 includes the above-described image decoding device.

The audio decoder 114 is, for example, an MPEG
In the case of / Audio Layer 1 / Layer 2, each of the electronic circuits includes a bit stream decomposition unit, an inverse quantization unit, and a subband synthesis filter bank unit. The input audio elementary stream is separated into a header, auxiliary information, and a quantized sub-band signal by a bit stream decomposing unit, and the quantized sub-band signal is dequantized by the number of bits allocated by the dequantizing unit. After being synthesized by the sub-band synthesis filter bank, it is output.

In such a camera-integrated recording / reproducing apparatus, a portion of high interest or important to the photographer or viewer can be designated by a color, and the reproduced image of the color region can be improved. For example, a flesh color can be specified, and when a flesh color is specified, the area of a person can be made to have high definition.

(Second Embodiment) The second embodiment is an image coding apparatus for improving the image quality by using not only the color of the pixel but also other features of the MB. Other features include, for example, activity, average luminance value, and residual value of the motion vector. The evaluation value is a combination of selecting two elements such as a combination of a pixel color and an activity from a pixel color, an activity, a luminance average value and a residual amount, and a pixel color, an activity and a luminance average value. There is a combination of selecting three elements like a combination and a combination of a pixel color, an activity, a luminance average value, and a residual amount. In the second embodiment, a pixel color, an activity, a luminance average This is a case where the evaluation value V (j) is created from the value and the residual amount.

FIG. 6 is a block diagram showing a configuration of an image coding apparatus according to the second embodiment.

In FIG. 6, the image encoding apparatus according to the second embodiment includes an image rearranging section 11, a subtracting section 12, a switch 1
3, DCT section 14, quantization section 15, variable length coding section 16,
Buffer memory 17, control unit 18, inverse quantization unit 21, inverse
DCT section 22, adder section 23, frame memories 24, 26,
Motion compensator 25, motion detector 27, residual calculator 28, frame complexity calculator 31, activity calculator 32, DC component calculator 33, and skin color component detector 34
It is comprised including. That is, the image encoding device according to the second embodiment is different from the image encoding device according to the first embodiment in that a residual operation unit 28, a frame complexity operation unit 31, an activity calculator 32 and a DC component calculator 33. For this reason, the configuration other than these different configurations is the same as that of the first embodiment, and therefore the description thereof will be omitted, and only the different configurations will be described below. In addition,
As in FIG. 1, without using the frame memory 26,
The case where each MV is calculated from the decoded image data in the frame memory 24 is indicated by a broken line.

The residual operation unit 28 determines the size of each MB of the P picture and the B picture based on the output of the motion detection unit 27.
Motion vector (MV)
Abbreviated. ) Is calculated. Let Bd (j) be the residual value for the j-th MB.

The activity calculation unit 32 calculates the activity representing the complexity of the image based on the luminance signal pixel value of the original image by using Expressions 14 to 16.

[0115]

(Equation 1)

[0116]

(Equation 2)

[0117]

(Equation 3)

Here, act (j) is the activity for the j-th MB. Pk is a pixel value in the luminance signal block of the original image. The reason for taking the minimum value in Equation 14 is to reduce the quantization scale when there is a flat portion even in only a part of the MB.

The DC component calculation unit 33 calculates the equation (17) for each MB.
, And outputs the average luminance value to the control unit 18.

[0120]

(Equation 4) Here, DC (j) is a luminance average value for the j-th MB, Yi
Is the luminance value for the i-th pixel in the MB when the luminance value is quantized in 256 steps.

While the activity calculation unit 32 calculates a value for each MB, the frame complexity calculation unit 31
Calculates the amount representing the complexity of the image for each frame by comparing the luminance between adjacent pixels.

The control unit 18 controls the skin color component detection unit 34
The evaluation value V (j) is obtained based on the output value h (j), the activity act (j), the residual value Bd (j), and the average luminance value DC (j).
The control unit 18 calculates a correction coefficient Nv (j) based on the evaluation value V (j).
And outputs it to the quantization unit 15 and the inverse quantization unit 21.

FIG. 7 is a flowchart for explaining main processing of the image coding apparatus according to the second embodiment.

In FIG. 7, the control unit 18 takes in the output h (j) of the skin color component detection unit 34 (St11). Next, the control unit 18 fetches the activity act (j) calculated by the activity calculation unit 31 (St12). Next, the control unit 18 determines whether or not the MB is an MB for performing the intra process (St13). As a result of the determination, in the case of the MB to be subjected to the intra process, the control unit 18 fetches the residual value Bd (j) calculated by the residual calculation unit 28 (St14), and performs the process of St15. On the other hand, as a result of the determination, if the MB is not subjected to the intra process, the control unit 18 performs the process of St15.

Next, the controller 18 takes in the DC (j) calculated by the DC component calculator 33 (St15).

Next, the control unit 18 generates an evaluation value V (j) of the MB based on Expression 18 (St16). When h (j) ≧ th_h, V (j) = act (j) × b1 (18-1) h (j) <th_h, and th_DClo ≦ DC (j) ≦ th_DCup, and Bd (j) When ≧ th_Bd, V (j) = act (j) × b2 (18-2) When h (j) <th_h, and when th_DClo ≦ DC (j) ≦ th_DCup, and when Bd (j) <th_Bd V (j) = act (j) × b3 (18-3) h (j) <th_h and DC (j)> th_DCup, or h (j) <th_h and DC (j) In the case of <th_DClo, V (j) = act (j) (18-4) (18)

Here, th_h is a threshold value for determining a skin color. b1, b2 and b3 are 0 in consideration of visual characteristics.
<B1 <b2 <b3 <1 Any constant that satisfies <1. That is, if MB is a skin color, the evaluation value V (j) is set to the smallest value b1 regardless of other conditions such as the activity (Equation 18-
1) When the residual is large in a certain brightness range, the evaluation value V (j) is set to the next smaller value b2 because the noise is conspicuous.
(Equation 18-2), and in the other range of the residual value, the evaluation value V (j) is set to b3 (Equation 18-3). For example, b1 = 0.5, b2
= 0.7 and b3 = 0.8.

The control unit 18 executes St11 to St16 for all MBs in the frame (St1
7). Next, the control unit 18 reads the generated bit amount (St18). Next, the control unit 18 sets a picture target bit amount for each picture, and sets a quantization scale reference value for each MB (St19).

Next, the control unit 18 generates a correction coefficient Nv (j) according to Equation 19 (St20). Nv (j) = (R × V (j) + avg_V) / (V (j) + R × avg_V) (19) where avg_V is the average value of V (j) in the picture, and R is The range of the correction coefficient Nv (j) is determined as in the first embodiment shown in Expression 7.

Next, the control section 18 determines the correction coefficient Nv
Using (j) in Expression 20, the quantization scale Q (j) for the MB of the subject is corrected (St21). Q (j) = refQ (j) × Nv (j) (20)

Next, the control unit 18 sets St20 and St
21 for all MBs in the frame (S
t22).

As described above, the image encoding apparatus according to the present embodiment detects the MB including the characteristic color of the subject of interest to the photographer, the viewer, and the like, thereby determining the subject in the entire image. I do. Then, the image encoding device allocates a quantization scale value for the detected MB in accordance with the visual characteristic using the correction coefficient obtained based on the visual characteristic, thereby setting the reproduction image quality of the specific subject to the first. It is more improved than the case of the embodiment.

[0133]

According to the present invention, it is possible to finely specify a portion to be quantized in an image by setting a specific color. According to the present invention, by setting, by color, a portion that the photographer or viewer is interested in or a portion that is considered important, such a portion can be quantized finely.
The image quality of important parts can be improved.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image encoding device according to a first embodiment.

FIG. 2 is a flowchart illustrating main processing of the image encoding device according to the first embodiment.

FIG. 3 is a diagram showing a relationship between a specific color and its area in a color difference signal space.

FIG. 4 is a block diagram illustrating a configuration of an image decoding device.

FIG. 5 is a block diagram showing a configuration of a camera-integrated digital recording / reproducing apparatus.

FIG. 6 is a block diagram illustrating a configuration of an image encoding device according to a second embodiment.

FIG. 7 is a flowchart illustrating main processing of an image encoding device according to a second embodiment.

[Explanation of symbols]

11 ... screen rearrangement unit, 12 ... subtraction unit, 13 ...
Switch, 14 ... discrete cosine transform unit, 15 ...
Quantization unit, 16: variable length coding unit, 17: buffer memory, 18: control unit, 21: inverse quantization unit, 22: inverse discrete cosine transform unit, 23 ... Addition units, 24, 26: Frame memory, 25: Motion compensation unit, 27: Motion detection unit, 28: Residual operation unit, 31: Frame complexity operation unit, 32
... Activity calculator, 33 ... DC component calculator, 34 ... Skin color component detector

 ──────────────────────────────────────────────────続 き Continued on the front page F-term (reference) 5C057 AA01 EA02 EA07 EF00 EG08 EL01 EM08 EM09 EM13 EM16 GG01 GH03 GH05 GJ09 GM01 5C059 KK02 MA00 MA22 MA23 MC11 MC38 ME02 NN01 PP16 SS14 SS20 TA46 TB07 TC10 TDTC TC TC TC

Claims (15)

[Claims] 1. An image coding apparatus for coding an original image for each block obtained by dividing the original image by a predetermined number of pixels, comprising: a color determination unit configured to determine whether a color of each block is a predetermined color; Quantization scale changing means for changing the quantization scale of each block according to the output of the color determination means, so that the quantization scale in the color block is smaller than the quantization scale in the block which is not the predetermined color, An image encoding device comprising: a quantization unit that quantizes an image of each block at each quantization scale determined by the quantization scale changing unit. 2. The quantization scale changing means further changes a quantization scale of each block so as to substantially maintain the amount of encoded data assigned to the original image. Item 2. The image encoding device according to Item 1. 3. The method according to claim 1, wherein the color determining unit determines a predetermined color for each pixel for each block, and further determines a ratio of a pixel of the predetermined color to the total number of pixels in the block. 2. The image coding apparatus according to claim 1, wherein the means further changes the quantization scale of the block of a predetermined color in accordance with the ratio such that the quantization scale decreases as the ratio increases. apparatus. 4. The image coding apparatus according to claim 1, wherein the color determination unit determines the color of the block using an average value of the color difference signals for all pixels in the block. 5. The quantization scale changing means further changes a quantization scale for a peripheral block of a block determined to be a predetermined color to a quantization scale smaller than a quantization scale of a block which is not a predetermined color. The image encoding device according to claim 1, wherein: 6. The quantization scale changing unit changes the quantization scale when changing the quantization scale and when the block is within a predetermined area on the screen. An image encoding device according to claim 1. 7. The image encoding apparatus according to claim 1, wherein the predetermined color is a flesh color. 8. An image processing apparatus further comprising: activity calculating means for obtaining an image activity; wherein the quantization scale changing means changes the quantization scale according to the activity from the activity calculating means when changing the quantization scale. The image encoding device according to claim 1, wherein 9. The image processing apparatus according to claim 1, further comprising: a brightness average value calculating unit that calculates a brightness average value of each block; wherein the quantization scale changing unit further includes a brightness average value from the brightness average value calculating unit when the quantization scale is changed. 2. The image encoding apparatus according to claim 1, wherein the quantization scale is changed according to the value. 10. A residual value calculating means for calculating a residual value of a motion vector in predicted image data, wherein the quantization scale changing means further changes the residual value when changing a quantization scale. 2. The image coding apparatus according to claim 1, wherein the quantization scale is changed according to a residual value from the image data. 11. An orthogonal transformation means for transforming an image divided into blocks each having a predetermined number of pixels into data suitable for encoding by two-dimensional orthogonal transformation for each block, and the color of each block is a predetermined color. Color determination means for determining whether or not each of the blocks has a predetermined color in accordance with an output of the color determination means such that a quantization scale in a block of a predetermined color is smaller than a quantization scale in a block which is not a predetermined color. Quantization scale changing means for changing a quantization scale; quantization means for quantizing an image of each block at each quantization scale determined by the quantization scale changing means; and an output of the quantization means being predetermined. And a variable-length coding unit for performing variable-length coding according to the obtained code table. 12. An image encoding method for encoding an original image for each block obtained by dividing the original image by a predetermined number of pixels, comprising: a color determination step of determining whether a color of each block is a predetermined color; A quantization scale changing step of changing the quantization scale of each block according to the result of the color determination step so that the quantization scale in the color block is smaller than the quantization scale in the block that is not the predetermined color; A quantization step of quantizing an image of each block with each quantization scale determined in the quantization scale changing step. 13. An image encoding program for causing a computer to encode an original image for each block obtained by dividing the original image by a predetermined number of pixels, a color determining step of determining whether a color of each block is a predetermined color, A quantization scale change that changes the quantization scale of each block according to the result of the color determination step so that the quantization scale of the block of the predetermined color is smaller than the quantization scale of the block that is not the predetermined color. And a quantization step of quantizing an image of each block at each quantization scale determined in the quantization scale changing step. 14. A computer-readable recording medium in which an image encoding program for causing a computer to encode an original image in blocks divided by a predetermined number of pixels is recorded, and whether a color of each block is a predetermined color or not. A color determination step of determining whether the quantization scale of a block of a predetermined color is smaller than a quantization scale of a block that is not a predetermined color. A quantization scale changing step of changing the quantization scale, and an image encoding program comprising a quantization step of quantizing an image of each block at each quantization scale determined in the quantization scale changing step. Characteristic recording medium. 15. An image capturing means for capturing an image of a subject, converting the image into a digital video signal and outputting the digital image signal, an image encoding means for compressively encoding image data from the image capturing means, and encoding by the image encoding means. A digital camera comprising: a recording unit that records encoded data on a recording medium, wherein the image encoding unit is the image encoding device according to claim 11.