JPH11234669A - Image compression processing unit and its method, and digital camera usign the unit - Google Patents

Abstract

PROBLEM TO BE SOLVED: To prevent image quality deterioration even at a high compression rate by updating compression parameters for each spatial frequency based on a calculation result of a space frequency distribution in an image so as to attain quantization corresponding to the characteristic of an image compression. SOLUTION: An analog image signal received from an input port 1 is converted into a digital signal by an image A/D converter 2 and an image signal processing circuit 3 applies a general gamma correction processing and a processing such as contour emphasis to the digital signal. A high frequency extraction circuit 6 detects a high frequency component with a high spatial frequency from the signal processed by the image signal processing circuit 3 and gives information to a compression control circuit 5. The compression control circuit 5 adjusts a quantization accuracy to keep a code amount rate within a prescribed range to compress the data so that a buffer in an image compression coder is not saturated. In the case of adjusting the code amount rate, not only a Q scale but also a Q table are changed. Thus, quantization accuracy is changed corresponding to an optical spatial frequency.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image compression processing apparatus and method, and more particularly, to an image compression processing apparatus and method suitable for use in digital equipment such as a digital camera, a DVD player, a digital TV receiver, etc., which performs image compression by a method such as MPEG. The present invention relates to an image compression processing device and method.

[0002]

2. Description of the Related Art In recent years, with the explosive spread of personal computers, it has become common to use image-based expression means for various communication media such as the Internet and intranets. This means that the video has started to be treated as one of information data, not just for viewing.

In such an environment, techniques such as video signal encoding and compression processing for efficient information transmission are indispensable. In particular, compression coding of a moving image requiring a high compression rate is performed by a combination of a plurality of algorithms such as DCT transform, inter-frame prediction, and motion compensation. As a result, efficient compression encoding is realized even for a subject including motion, using the correlation between adjacent frame images and the motion information of the subject. These encoding methods include H.264, which is an encoding standard for videophones and communications. H.261 and MPEG which is a moving image storage encoding method are generally known.
Regarding these encoding techniques, see Journal of the Institute of Television Engineers of Japan, V
ol. 45, No7, pp. 793-799 (199
1), and the same pp. 807-812 and the like.

[0004]

One of the most important techniques among the data compression techniques found in the above prior art is control of the code rate. This is a control for averaging the data rate in consideration of not causing a failure in the buffer on the decoder side.

The control for equalizing the code amount is controlled by changing the quantization coefficient.

The quantization coefficient is a coefficient used when quantizing data arranged and encoded in a spatial frequency component by DCT or the like. In general, a quantization circuit (hereinafter referred to as a “Q table”) is provided with a reference value that defines a quantization coefficient for each spatial frequency in a quantization circuit, and is used to increase or decrease the generated code amount after compression. Accordingly, each value of the quantization table is converted to a quantization scale (hereinafter, “Q scale”).
Is multiplied by a constant value called as a quantization coefficient.

The actual quantization is performed by dividing by this quantization coefficient (= Q scale × Q table value) and rounding the remainder.

Hereinafter, this will be described in some detail using mathematical expressions.

[0009] MPEG / ISO / IEC 11172-2 D
According to 6.3.4, the quantization result value is determined by the following equation.

[0010]

(Equation 1)

Here, [u, v] is an argument indicating a position in the DCT block which is a basic unit of compression, and the larger this numerical value, the higher the spatial frequency.

The operation symbol // means that the division result is rounded off.

The actual encoding is performed based on the result of the quantization performed as described above. Here, when lowering the quantization precision, that is, when it is desired to reduce the amount of generated code, the quantization coefficient of the denominator of the above (Equation 1) is increased to reduce the variation of the quantization result value, Many “0” s appear in the result value. By doing so, the compression rate in encoding is improved, a high compression rate is obtained, and the amount of generated code is reduced.
However, when the compression rate is increased and the code amount is reduced, an error with the actual image data increases, so that the image quality deteriorates. The biggest problem of the image compression algorithm is to reduce the code amount while keeping the deterioration of the image quality within an allowable range.

Conversely, when the quantization precision is to be increased, that is, when it is desired to increase the amount of codes to be generated, (Equation 1)
Reduce the quantization coefficient of the denominator of. At this time, the compression rate in encoding is reduced and the compression rate is reduced, so that the generated code amount increases. At this time, the quality of the image is relatively improved.

Next, quantization will be described using a specific example of the Q table shown in FIG. Figure 7 shows the ISO / IEC
FIG. 11172-2 is a diagram showing a recommended example of a Q table applied to an Intra image described in section 2.4.3.2.

When the above (Equation 1) is applied to this table, the Q table value of the DC component ([u, v] = [0, 0]) is 8, while the component having the highest spatial frequency (8) is obtained. The Q table value of [u, v] = [7, 7]) is 83, which indicates that the value of the quantization result decreases as the spatial frequency increases even when there are equal amounts of data values. Is shown. The significance of quantizing with a table having a gradient depending on the spatial frequency in this way is to consider that the higher the spatial frequency, the lower the human visual sensitivity is,
An object of the present invention is to perform more efficient compression such that the apparent image quality does not change so much even when the compression ratio is increased.

Next, the relationship between the quantization accuracy and the code rate will be described with reference to FIG. FIG. 8 is a configuration diagram of an MPEG image compression apparatus according to the configuration of the related art.

The input image data is orthogonally transformed by the DCT circuit 501 and sent to the quantization circuit 502. The quantization circuit performs a quantization process. That is,
As shown in (Equation 1), a process of reducing the code amount by dividing the entire code amount by the quantization coefficient and expressing it by a small numerical value is performed. The signal after quantization is sent to the motion vector compensation circuit 503 and fed back so that compression suitable for moving image compression can be performed. The motion vector compensation is a function for performing efficient compression using temporal correlation of images.

The quantized signal is supplied to an encoding circuit 50.
In 4, processing such as hierarchical structure encoding or variable length encoding is performed and transmitted to the transmission buffer 505. The data stored in this buffer is the final amount of data (code) after compression.

On the other hand, the amount of data accumulated in the transmission buffer is constantly monitored, and this information is also transmitted to a quantization circuit and, in some cases, an encoding circuit, fed back, and transmitted per unit time of the transmitted code. The amount, ie, the code rate, is controlled.

In the quantization circuit, when the code rate is larger than the allowable code rate, the quantization coefficient of (Equation 1) is increased and the quantization precision is reduced (that is, as high compression).
It works to reduce the final data amount, and in the opposite case, the quantization coefficient of (Equation 1) is reduced, and the quantization precision is increased.

As described above, in an apparatus that performs image compression by MPEG, the code rate can be controlled by changing the quantization precision performed by the quantization circuit. By the way, in general, the spatial frequency of a subject is naturally not uniform. For example, there are some subjects whose spatial frequencies are concentrated in the middle and low frequencies, and some subjects are concentrated only in the high frequencies. In such a case, the Q table that is optimal for quantization, that is, that can achieve the most effective compression also changes.

However, in the general code amount control according to the prior art, each coefficient on a predetermined Q table is simply changed to a constant value (Q scale) according to an increase or decrease of the generated code amount. Multiplication is performed to obtain a quantized coefficient, and the data value is divided by the quantized coefficient to obtain a quantized result. Therefore, quantization according to the frequency characteristics of the code cannot be performed, and there is a problem that it is difficult to always maintain optimal coding.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to make it possible to perform quantization in accordance with the frequency characteristics of an image when performing quantization for image compression. Another object of the present invention is to provide an image compression processing apparatus and method which do not deteriorate image quality even at a high compression ratio.

Another object of the present invention is to provide a digital camera having an autofocus function to which the image compression processing method is applied.

[0026]

In order to achieve the above object, an image compression processing apparatus according to the present invention comprises an image compression processing apparatus for converting an image into a digital signal and performing compression encoding. An image signal processing unit for generating a digital image signal from an input signal, an image compression encoding unit for compressing and encoding the digital image signal, and setting parameters relating to image compression for the image compression encoding unit, And a high-frequency detecting means for detecting a high-frequency component contained in the digital signal, wherein the compression controlling means controls a spatial frequency in a screen based on an output result of the high-frequency detecting device. The distribution is calculated, and based on the calculation result, the update of the compression parameter given to the image compression encoding unit is updated for each spatial frequency. It is obtained as performed in.

More specifically, in the image compression processing apparatus, the compression parameters are a quantization scale and a quantization table, and the value of the corresponding quantization table can be changed for each spatial frequency. It was made.

More specifically, in the image compression processing apparatus, when there is a maximum value of the spatial frequency energy distribution of the digital signal, the value of the quantization table corresponding to the frequency having the maximum value is increased. The image compression is performed by reducing the quantization precision of the digital data of the frequency.

To achieve the above object, an image compression processing method according to the present invention comprises an image compression processing method for converting an image into a digital signal and performing compression encoding. The processing device includes an image signal processing unit that generates a digital image signal from the input signal, an image compression encoding unit that compresses and encodes the digital image signal, and a parameter setting related to image compression for the image compression encoding unit. And a compression control means for giving a command, and further comprising a high frequency detection means for detecting a high frequency component included in the digital signal, wherein the compression control means , And based on the calculation result, further updates the compression parameters given to the image compression encoding means. The, in which as performed for each spatial frequency.

More specifically, in the above image compression processing method, the compression parameters are a quantization scale and a quantization table, and the value of the corresponding quantization table can be changed for each spatial frequency. It was made.

More specifically, in the above image compression processing method, when there is a maximum value of the energy distribution of the spatial frequency of the digital signal, the value of the quantization table corresponding to the frequency having the maximum value is increased. The image compression is performed by reducing the quantization precision of the digital data of the frequency.

In order to achieve the above object, a digital camera according to the present invention comprises a digital camera having a function of converting a subject into a digital signal, performing compression coding, and recording the digital signal. Image signal processing means for generating a digital image signal, image compression / encoding means for compression-encoding the digital image signal, and compression control for setting an image compression parameter for the image compression / encoding means and giving a command. Means,
Furthermore, a lens, an image sensor, and an automatic focusing (autofocus) unit are provided, and the automatic focusing unit has a function of detecting a high-frequency component included in a subject image. The spatial frequency distribution in the screen is calculated based on the output result of the high-frequency component detection detected by the automatic focusing unit. Based on the calculation result, the image compression encoding unit is subjected to a compression parameter for each spatial frequency. Is updated, compression encoding is performed, and a subject image is recorded.

[0033]

DESCRIPTION OF THE PREFERRED EMBODIMENTS Embodiments of the present invention will be described below with reference to FIGS. [Embodiment 1] Hereinafter, a first embodiment of the present invention will be described with reference to FIG.
This will be described with reference to FIG. First, the configuration of the image compression processing device according to the present invention will be described with reference to FIG. FIG. 1 is a configuration diagram of an image compression processing device according to the present invention.

The image compression processing apparatus according to the present invention
It is used when encoding a still image or a moving image by a method such as G, and is built in a digital camera or embodied as an MPEG board of a personal computer.

First, an analog image signal is input from the input port 1 and converted into a digital signal by the image A / D converter 2. This digital signal is supplied to the image signal processing circuit 3
Thus, general gamma correction processing, contour emphasis processing, and the like are performed. The image signal processing circuit 3 is a circuit generally called a DSP (Digital Signal Processor).

The image compression encoder 4 encodes and compresses an image signal according to algorithms such as DCT transform, quantization, frame prediction, and motion compensation. Further, the image encoder 4 converts the compressed data into MPEG data.
A bit stream is generated in a form including various kinds of coding information in accordance with standards such as

The high frequency extraction circuit 6 detects a high frequency component having a high spatial frequency from the signal processed by the image signal processing circuit 3 and transmits the information to the compression control circuit 5.

The compression control circuit 5 compresses the data by adjusting the quantization precision so as to keep the code rate within a certain range so that the buffer inside the image compression code does not become saturated.

Next, the principle of the image compression method of the present invention will be described with reference to FIGS. FIG. 2 is a conceptual diagram showing the principle of the adjustment of the quantization accuracy of the related art and the principle of the adjustment of the quantization accuracy of the present invention in comparison. Here, the vertical axis represents the quantization accuracy, and the horizontal axis represents the horizontal and vertical spatial frequencies obtained by the DCT transform. FIG. 3 is a diagram illustrating an example of an image taken up in the embodiment. FIG.
Is a graph showing the energy distribution that can be taken for each spatial frequency by adjusting the quantization accuracy, in comparison with the prior art and the present invention.

As described in the section of the prior art, when controlling the code rate, only the Q scale is changed in order to adjust the quantization precision. This will be described with reference to the principle diagram of the three-dimensional graph in FIG. 2, in which a predetermined quantization table is moved in parallel with respect to an axis of quantization accuracy. As a result, when the generated code amount is larger than the allowed fixed amount and it is desired to decrease the amount, the quantization accuracy is lowered, and conversely, since the generated code amount is smaller than the allowed amount, the quantization accuracy is increased. If it is desired, the quantization accuracy should be increased. By manipulating the quantization precision in this way, the code rate can be made uniform. Specifically, as shown in (Equation 1), increasing the quantization accuracy means reducing the value of the Q scale, and decreasing the quantization accuracy means increasing the value of the Q scale. It is to be.

On the contrary, in the present invention, not only the Q scale but also the Q table is changed to adjust the quantization precision. In this way, the quantization precision can be changed corresponding to an arbitrary spatial frequency as shown in FIG. That is, in the principle of the prior art of FIG. 2A, the adjustment of the quantization precision is linear, but in the present invention, the value of the quantization table can be changed, and as shown in FIG. Non-linear adjustments can also be made. That is, specifically, the quantization coefficient (= Q scale × Q table) in the denominator of (Equation 1) is changed. When the quantization precision is increased, the value of the quantization coefficient is reduced, When lowering the quantization precision, the value of the quantization coefficient is increased.

In the example shown in FIG. 2B, the code rate is reduced by lowering the quantization precision, particularly where the spatial frequency is higher than a certain spatial frequency.

According to the present invention, based on such a principle, as a means for realizing the actual circuit, as shown in FIG.
Is such that both values of the Q scale and the Q table can be designated.

Next, the image compression method of the present invention will be described in detail based on an actual image with reference to FIGS.

For example, consider the case where a person wearing a shirt with fine stripes is imaged as shown in FIG. In this case, the striped portion β of the shirt increases the code amount,
The Q scale increases. For this reason, in the related art, when the quantization precision is adjusted to keep the code amount constant, the quantization precision in all spatial frequencies decreases, and large block distortion appears in the face portion α of the person, Or mosquito noise appears at the boundary with.

This is because, in a case where information about a certain frequency is specifically present in a certain subject and the code amount is increased due to the information, for example, the control of the conventional quantization scale alone does not cover all spatial frequencies. This is because the generated code amount can be reduced only by uniformly lowering the quantization accuracy.

On the other hand, in the adjustment of the quantization precision of the present invention, Q
Not only the scale but also the Q table can be changed for each frequency.

That is, in this example, attention is paid to the vicinity of the spatial frequency of the striped portion β of the shirt, and the quantization precision corresponding to this frequency is particularly reduced. In such a case, only the accuracy of the expression of the stripes on the shirt is reduced, but the face portion α
The impact on such as will be reduced.

This will be specifically described with reference to the graph sequence of FIG.

FIG. 4A shows the code energy distribution of a subject in this example, taking the spatial frequency of the subject on the horizontal axis and the code amount on the vertical axis. The low frequency corresponds to the face portion α, and the high frequency corresponds to the portion β of the shirt stripe, and energy (code amount) is concentrated in this portion. That is, the portion corresponding to the stripe portion β of the shirt protrudes, and has a local maximum value larger than that of the vicinity.

FIG. 4B1 shows the values of the quantization table for each frequency component according to the prior art, which has a constant gradient. Note that, in practice, a discrete value in a step shape is taken, but for the sake of simplicity, the value is changed continuously.

On the other hand, FIG. 4B2 shows the values of the quantization table for each frequency component according to the present invention, and the values of the quantization table corresponding to the striped portion β of the shirt are large.

In adjusting the quantization precision according to the prior art,
In the code amount compressed by (Equation 1), a portion corresponding to the stripe portion β of the shirt protrudes as shown in FIG. 4C1. In the present invention, on the other hand, the portion corresponding to the striped portion β of the shirt becomes gentle as shown in FIG. Therefore, the code amount of the low frequency portion can be relatively increased, and large block distortion appears in the face portion α,
Mosquito noise can be prevented from appearing at the boundary with the background.

As described above, according to the present invention, it is possible to perform finer-grained encoding control in accordance with the situation of a subject, and it is possible to achieve an overall improvement in image quality.

Next, the structure of the high-frequency extraction circuit 6 will be described with reference to FIG. FIG. 5 is a block diagram showing the structure of the high-frequency detection circuit of the present invention.

Here, as shown in FIG. 5, the high-frequency extraction circuit 6 includes a plurality of high-pass filter band-pass filters 601 and an area designating circuit 6 for designating an area in the screen.
02 and an adder circuit 603. In the simplest case, the area specification circuit specifies the entire screen,
All signals in this area are sent to the filter.
The filter output signal is stored in the adding circuit 603. If the adder circuit is reset at the start of one-screen scanning, the adder indicates at the end of one-screen scanning that the total amount of the high- or mid-range components that have passed through the respective filters present on that screen. Since the data is contained, it is possible to estimate the spatial frequency distribution of the image by verifying the data.

[Embodiment 2] Next, a second embodiment of the present invention will be described with reference to FIG. FIG. 6 is a block diagram showing the structure of a digital camera using the image compression processing of the present invention.

This digital camera first has a lens 30
In step 1, an optical signal of a subject image is formed on an image sensor 302 such as a CCD. The image pickup device 302 converts the formed optical signal into an analog electric signal, and then converts the image A / D converter 30
The signal is converted into a digital signal by 3 and input to the image signal processing circuit 304. The luminance signal is branched and input to the high-frequency extraction circuit 306. The high-frequency extraction circuit is shown in FIG.
As shown in FIG. 5, the image processing apparatus includes a single or a plurality of bypass filters, band-pass filters, and the like. From signals obtained from these filters, image high-frequency information 308 indicating high-frequency information existing in the subject is obtained. It is generated and output to the automatic focusing controller 308.

The automatic focusing controller 308 is generally a microcomputer chip (Micro Controller Chip), and controls the focus motor 309 in accordance with high frequency information. In other words, to focus the camera means that the image has a high frequency, and the focus motor 309 is controlled to adjust the position of the lens so that the image is shifted toward the high frequency.

On the other hand, the image information output from the image signal processing circuit 304 is
The data is compressed and encoded in the EG format or the like. In the encoding, an encoding coefficient setting instruction 311 from the compression controller 310
The compression accuracy is determined based on

Here, in the present invention, the image high-frequency information 307 extracted by the high-frequency extraction circuit 306 is branched and input to the compression controller 310 as well.

The compression controller controls the input image high frequency information 3
07 is analyzed, and based on this, the Q scale and the Q scale are applied in accordance with the principle of the image compression processing of the present invention described in the first embodiment.
An instruction to update one or both of the table values is output to the image compression encoder 305.

In this way, it is possible to realize the image compression processing according to the present invention which can perform finer coding control in accordance with the situation of the subject while utilizing the circuit of the conventional autofocus function. it can.

[0064]

According to the present invention, it is possible to perform quantization in accordance with the frequency characteristics of an image at the time of quantization for compressing an image, and to perform image compression that does not degrade image quality even at a high compression rate. A processing apparatus and method can be provided.

Further, according to the present invention, it is possible to provide a digital camera having an autofocus function to which the image compression processing method is applied.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an image compression processing device according to the present invention.

FIG. 2 is a conceptual diagram showing the principle of adjustment of the quantization accuracy according to the prior art and the principle of the adjustment of the quantization accuracy according to the present invention.

FIG. 3 is a diagram illustrating an example of an image taken up in the embodiment.

FIG. 4 is a graph showing the energy distribution that can be obtained for each spatial frequency by adjusting the quantization accuracy, in comparison with the related art and the present invention.

FIG. 5 is a block diagram illustrating a structure of a high-frequency detection circuit according to the present invention.

FIG. 6 is a block diagram showing the structure of a digital camera using the image compression processing of the present invention.

[Fig. 7] ISO / IEC 11172-2 I described in section 2.4.3.2
FIG. 14 is a diagram illustrating a recommended example of a Q table applied to an ntra image.

FIG. 8 is a configuration diagram of an MPEG image compression device according to a configuration of the related art.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Input port 2 ... Image A / D converter 3 ... Image signal processing circuit 4 ... Image compression encoder 5 ... Code amount control circuit 6 ... Image high frequency extraction circuit 7 ... Output port 301 ... Lens 302 ... Image sensor 303 ... AD converter 304 image signal processing circuit 305 image compression encoder 306 high-frequency extraction circuit 307 image high-frequency information 308 automatic focusing (autofocus) controller 309 focus motor 310 compression controller 311 code Coefficient setting instruction

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Hiroaki Kami 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside Hitachi Image Information System Co., Ltd. (72) Hiroshi Sakurai 292 Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside Hitachi Image Information System Co., Ltd. (72) Koji Ooi, Inventor Koji 292, Yoshida-cho, Totsuka-ku, Yokohama-shi, Kanagawa Prefecture Inside Company Hitachi Image Information System Co., Ltd. (72) Hideto Noguchi 1410 Inada, Hitachinaka-shi, Ibaraki Stock Association Hitachi, Ltd. Video Information Media Division (72) Inventor Atsuhiro Ono 1410 Inada, Hitachinaka-shi, Ibaraki Prefecture Co., Ltd. Hitachi Ltd. Video Information Media Division (72) Inventor Kazunori Uemura 1410 Inada, Hitachinaka-shi, Ibaraki Hitachi, Ltd. Inside the Visual Information Media Division of Manufacturing

Claims (7)

[Claims] 1. An image compression processing device for converting an image into a digital signal and performing compression encoding, comprising: an image signal processing means for generating a digital image signal from an input signal; and an image for compressing and encoding the digital image signal. Compression encoding means; compression control means for setting parameters relating to image compression to the image compression encoding means and giving a command; and high frequency detection means for detecting a high frequency component contained in the digital signal. The compression control means calculates a spatial frequency distribution in a screen based on the output result of the high-frequency detection device, and, based on the calculation result, a compression parameter given to the image compression encoding means. An image compression processing device for updating the image data for each spatial frequency. 2. The method according to claim 1, wherein the compression parameters are a quantization scale and a quantization table, and a value of the quantization table corresponding to each spatial frequency can be changed. Image compression processing device. 3. When there is a maximum value of the energy distribution of the spatial frequency of the digital signal, the value of the quantization table corresponding to the frequency having the maximum value is increased, and the quantization accuracy of the digital data at that frequency is increased. 3. The image compression processing apparatus according to claim 2, wherein the image compression is performed by reducing the size. 4. An image compression processing method for converting an image into a digital signal and performing compression encoding, wherein the image compression processing device used for the method comprises: an image signal processing means for generating a digital image signal from an input signal; Image compression encoding means for compressing and encoding the digital image signal, and compression control means for setting a parameter relating to image compression to the image compression / encoding means and giving a command; further included in the digital signal High-frequency detection means for detecting a high-frequency component to be detected, the compression control means calculates a spatial frequency distribution in a screen based on an output result of the high-frequency detection device, and based on the calculation result, the image compression code Image compression processing method for updating a compression parameter given to a converting means for each spatial frequency 5. The method according to claim 4, wherein the compression parameters are a quantization scale and a quantization table, and the value of the quantization table corresponding to each spatial frequency can be changed. Image compression processing method. 6. When there is a maximum value of the energy distribution of the spatial frequency of the digital signal, the value of the quantization table corresponding to the frequency having the maximum value is increased, and the quantization accuracy of the digital data at that frequency is increased. 6. The image compression processing method according to claim 5, wherein the image compression is performed with the size reduced. 7. A digital camera having a function of converting a subject into a digital signal, performing compression encoding, and recording the image, an image signal processing unit for generating a digital image signal from a subject image, and compressing the digital image signal. An image compression encoding means for encoding, and a compression control means for setting parameters relating to image compression to the image compression encoding means and giving a command; and further comprising a lens, an image sensor, and an automatic focusing device. (Autofocus) means, wherein the automatic focusing means has a function of detecting high-frequency components contained in the subject image, and the compression control means outputs the high-frequency component detection detected by the automatic focusing means. The spatial frequency distribution in the screen is calculated based on the result, and based on the calculated result, the image compression encoding means is provided for each spatial frequency, Perform the update of the condensation parameters, digital camera and recording the subject image compression coding.