JP6191173B2 - Image processing apparatus, method, and program - Google Patents

Description

  The present invention relates to an image compression technique in an image processing apparatus.

  Various image processing apparatuses use a compression technique such as JPEG (Joint Photographic Experts Group).

  Patent Document 1 discloses a compression code for image data for the purpose of increasing the accuracy and speed of code amount control with a simple configuration in an image data processing apparatus that compresses and encodes image data by two-dimensional orthogonal transformation. A configuration is disclosed in which preliminary compression encoding is performed to obtain appropriate bit allocation prior to conversion.

  Patent Document 2 discloses an image processing apparatus having a compression processing function for image data, and performs frequency conversion of an input image in units of blocks composed of a plurality of pixels for the purpose of improving compression efficiency and editability of compressed data. Means for outputting a plurality of frequency transform coefficients, means for quantizing a plurality of frequency transform coefficients and outputting a plurality of quantization coefficients, and limiting a range of values of the plurality of quantization coefficients or frequency transform coefficients The structure provided with a means is disclosed.

  The JPEG compression method has a feature that the code amount (coding rate) does not become constant according to the feature of the input image. For example, even if two different images are compressed with the same parameter, the code amount is not the same. For this reason, in order to control the code amount to a desired value, pre-compression may be performed prior to the main compression to determine parameters used during the main compression.

  The above-mentioned patent document 1 is an example when the preliminary compression is performed. However, the preliminary compression in the related art is premised on the existence of a camera processing system circuit for performing activity calculation and the like. For this reason, there is a problem that prediction control by sufficient preliminary compression cannot be performed only with basic components necessary for compression processing such as JPEG.

  Therefore, an object of the present invention is to realize highly accurate code amount control using basic components necessary for compression processing.

In order to solve the above problems and achieve the object, an image processing apparatus according to the present invention includes an image dividing unit that divides an input image into a plurality of regions, and quantizes the divided input image based on a quantization table. Quantization means for encoding, encoding means for encoding the quantized input image, table holding means for holding a plurality of preliminary compression quantization tables created in advance for different quality values, and at the time of preliminary compression And a table control means for causing the quantization means to perform quantization of the input image by applying the pre-compression quantization table different for each region, and a code amount for detecting a code amount output by the encoding means Based on the detection means and the code amount for each preliminary compression quantization table detected during the preliminary compression and the target code amount during the main compression, the encoding during the main compression is performed. As the amount of codes stage output becomes the target code amount, but and a table determining means for determining the quantization table used during the main compression.

  According to the present invention, high-accuracy code amount control can be realized using basic components necessary for compression processing.

FIG. 1 is a diagram illustrating a configuration of a main part of the image processing apparatus according to the present embodiment. FIG. 2 is a diagram illustrating a specific configuration of the image processing apparatus according to the present embodiment. FIG. 3 is a diagram illustrating an input image divided by the preprocessing circuit shown in FIG. FIG. 4 exemplifies the preliminary compression quantization table A, and is a diagram showing a luminance / color difference quantization table when the quality value is 90. FIG. 5 exemplifies the preliminary compression quantization table B, and is a diagram showing a luminance / color difference quantization table when the quality value is 100. FIG. 6 exemplifies the preliminary compression quantization table C, and is a diagram showing a luminance / color difference quantization table when the quality value is 60. FIG. 7 exemplifies the preliminary compression quantization table D, and is a diagram showing a luminance / color difference quantization table when the quality value is 80. FIG. 8 is a diagram illustrating characteristic lines indicating the relationship between the Q value and the code amount in JPEG. FIG. 9 is a flowchart showing processing when determining the compression quantization table in the image processing apparatus according to the present embodiment. FIG. 10 is a flowchart showing processing executed when the target code amount is equal to or larger than the code amount corresponding to the preliminary compression quantization table A and equal to or smaller than the code amount corresponding to the table B in the processing shown in FIG. FIG. 11 is a flowchart showing processing executed when the target code amount is equal to or larger than the code amount corresponding to the preliminary compression quantization table C and equal to or smaller than the code amount corresponding to the table D in the processing shown in FIG. FIG. 12 is a flowchart showing processing executed when the target code amount is larger than the code amount corresponding to the preliminary compression quantization table D and smaller than the code amount corresponding to the table A in the processing shown in FIG.

  Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 shows a configuration of a main part of an image processing apparatus 1 according to the present embodiment. The image processing apparatus 1 has a function of compressing input image data by a method such as JPEG and performing a preliminary compression prior to the main compression. The preliminary compression according to the present embodiment is a process performed to optimize the quantization process so that the code amount output in the main compression matches a predetermined target code amount.

  The image processing apparatus 1 includes an image division unit 2, a quantization unit 3, an encoding unit 4, a table holding unit 5, a table control unit 6, a code amount detection unit 7, and a table determination unit 8.

  The image dividing unit 2 divides the input image into a plurality of areas. The quantization unit 3 quantizes the input image divided by the image dividing unit 2 based on the quantization table. The encoding unit 4 encodes (compresses) the input image quantized by the quantization unit 3.

  The table holding unit 5 holds a plurality of preliminary compression quantization tables (to be described in detail later) that are used for preliminary compression and created for different quality values. At the time of preliminary compression, the table control unit 6 applies different preliminary compression quantization tables for each region of the input image, and causes the quantization unit 3 to perform quantization of the input image.

  The code amount detection unit 7 detects the code amount of data output when the encoding unit 4 encodes the input image. Based on the code amount of each pre-compression quantization table detected during the pre-compression and the target code amount during the main compression detected by the code amount detection unit 7, the table determining unit 8 performs the encoding unit 4 during the main compression. The quantization table to be used at the time of the main compression is determined so that the code amount output by the is the target code amount.

  Each of the components 2 to 8 may be realized by software, for example, by causing a processing device such as a CPU (Central Processing Unit) to execute a program, that is, realized by software or hardware such as an IC (Integrated Circuit). Alternatively, software and hardware may be used in combination.

  According to the image processing apparatus 1, input image data supplied from the outside is processed by the image dividing unit 2 so that the input image is divided into a plurality of regions, and then the quantization unit 3 performs a predetermined quantization table. Quantized based on At this time, in the preliminary compression, the quantization unit 3 performs quantization by applying a plurality of different preliminary compression quantization tables held by the table holding unit 5 for each area of the input image. The input image data quantized in the preliminary compression thus obtained is encoded by the encoding unit 4. In this preliminary compression, the code amount of data output from the encoding unit 4 is detected by the code amount detection unit 7 and supplied to the table determination unit 8. The table determination unit 8 analyzes the supplied code amount at the time of preliminary compression for each type of the pre-compression quantization table, and compares the code amount for each pre-compression quantization table with the target code amount at the time of main compression. Then, based on the comparison result, the main compression quantization table for matching the code amount in the main compression with the target code amount is determined. Then, the table control unit 6 applies the main compression quantization table determined by the table determination unit 8 to the quantization in the main compression.

  According to the image processing apparatus 1, it is possible to perform preliminary compression for determining an optimum quantization table to be used at the time of main compression by only basic components necessary for normal compression processing. As a result, it is possible to realize highly accurate code amount control using basic components necessary for compression processing.

  FIG. 2 illustrates a specific configuration of the image processing apparatus 1 according to the present embodiment. The image processing apparatus 1 according to this example employs the JPEG method, and includes a preprocessing unit 11, a preprocessing memory 12, a DCT (Discrete Cosine Transform) unit 13, a quantization unit 14, a Huffman encoding unit 15, Code output control unit 16, code amount counter 17, main compression table determination unit 18, main compression table database (DB) 19, target code amount holding memory 20, table switching control unit 21, preliminary compression table database 22, and book A compression table holding memory 23 is provided.

  The pre-processing unit 11 is a form of the image dividing unit 2 in FIG. 1 and is a circuit or the like that performs processing before performing quantization and encoding on input image data to be compressed. The preprocessing includes raster / block conversion, the above-described input image division processing, and the like. The input image data is temporarily stored in the preprocessing memory 12. The preprocessing memory 12 functions as a work area for the preprocessing unit 11. The DCT unit 13 is a circuit that performs a known discrete cosine transform.

  FIG. 3 illustrates the input image 25 divided by the preprocessing unit 11. The input image 25 is divided in units of horizontal pixels and vertical lines. The division size is arbitrary, but the minimum value is a macroblock unit which is a JPEG processing unit. For example, in 4: 4: 4 sub-sampling, the 8 × 8 pixel unit is the minimum value. The finer the division size, the higher the accuracy of code amount control in main compression (prediction accuracy in preliminary compression), but the calculation load increases. As a general rule, the division size is a multiple of a macroblock unit. The input image 25 according to this example is divided into 8 × 8 = 64 areas.

  The quantization unit 14 is divided into a plurality of regions by the pre-processing unit 11 and pre-compression quantization tables A and B output from the table switching control unit 21 for the input image data subjected to discrete cosine transform by the DCT unit 13. , C, D, or a circuit that performs quantization using the main compression quantization table.

  The table switching control unit 21 is a form of the table control unit 6 in FIG. 1 and is a circuit or the like that switches the quantization table output to the quantization unit 14. The table switching control unit 21 outputs the preliminary compression quantization tables A, B, C, and D held in the preliminary compression table database 22 during preliminary compression, and performs preliminary compression in units of areas of the input image 25 shown in FIG. The quantization tables A, B, C, and D are switched. The table switching control unit 21 outputs the main compression quantization table held in the main compression table holding memory 23 during the main compression.

  FIG. 4 exemplifies the preliminary compression quantization table A, and shows the luminance / color difference quantization table when the quality value (Q value) is 90. FIG. 5 exemplifies the preliminary compression quantization table B, and shows the luminance / color difference quantization table when the Q value is 100. FIG. 6 exemplifies the preliminary compression quantization table C, and shows the luminance / color difference quantization table when the Q value is 60. FIG. 7 exemplifies the preliminary compression quantization table D, and shows the luminance / color difference quantization table when the Q value is 80. In each of FIGS. 4 to 7, the left table represents the luminance, and the right table represents the color difference table. The larger the Q value, the smaller the compression rate and the smaller the image degradation. The smaller the Q value, the greater the compression ratio and the greater the degradation of the image.

  In FIG. 3, “A area”, “B area”, “C area”, and “D area” are included in each area of the input image 25, respectively. It shows that the quantization table B, the preliminary compression quantization table C, and the preliminary compression quantization table D are applied to perform quantization.

  FIG. 8 exemplifies characteristic lines indicating the correlation between the Q value and the code amount in JPEG. In the case of such characteristics, for example, when it is desired to control the code amount with a Q value of 50 or more, the preliminary compression quantization table AB (90 ≦ Q value ≦ 100), the preliminary compression quantization table By dividing the characteristics into three regions of CD (60 ≦ Q value ≦ 80) or preliminary compression quantization table DA (80 <Q value <90), prediction by preliminary compression can be performed well. it can.

  Since such characteristic lines change depending on the type of input image and the like, for example, a plurality of characteristic lines for natural images, for humans, for documents (characters), and corresponding precompression quantizations. It is also possible to prepare a table. It is also possible to perform selection according to the type of input image according to user operation, appropriate automatic image recognition processing, or the like.

  The Huffman encoder 15 (see FIG. 2) is a circuit that encodes the input image data quantized by the quantizer 14 based on a known Huffman encoding algorithm. The code output control unit 16 is a circuit or the like that selects an output destination of Huffman-encoded input image data. The code output control unit 16 outputs the encoded data to the code amount counter 17 at the time of preliminary compression, and outputs the encoded data to the predetermined device such as an external device as final compressed data at the time of the main compression.

  The code amount counter 17 is a circuit that counts the code amount of the data output from the code output control unit 16. The code amount counter 17 counts the code amount for each of the preliminary compression quantization tables A, B, C, and D. In this example, 1/4 of the code amount of one input image 25 is counted for each of the preliminary compression quantization tables A, B, C, and D. That is, the code amount four times the recorded code amount becomes the code amount of each of the preliminary compression quantization tables A, B, C, and D.

  The main compression table determination unit 18 includes a code amount for each of the preliminary compression quantization tables A, B, C, and D output from the code amount counter 17 and a target code amount in main compression held in the target code amount holding memory 20. And a circuit for performing a process of determining a main compression quantization table to be used in the main compression. The main compression table determination unit 18 calculates the Q value of the main compression quantization table, for example, based on the comparison result between the code amount of each of the preliminary compression quantization tables A, B, C, and D and the target code amount. Then, the main compression quantization table corresponding to the Q value is extracted from the main compression table database 19. The extracted main compression quantization table is held in the main compression table holding memory 23.

  Then, the table switching control unit 21 outputs the main compression quantization table held in the main compression table holding memory 23 to the quantization unit 14 at the time of main compression. Thereby, it is possible to perform quantization using the optimum main compression quantization table corresponding to the target code amount.

  2 may be realized by software by causing a processing device such as a CPU (Central Processing Unit) to execute a program, that is, an IC (Integrated Circuit) or the like. It may be realized by hardware, or may be realized by using software and hardware together.

  When the software is used, the control program and the like are preferably provided by being incorporated in advance in a ROM (Read Only Memory). The control program is a file in an installable or executable format and is recorded on a computer-readable recording medium such as a CD-ROM, a flexible disk (FD), a CD-R, or a DVD (Digital Versatile Disk). And may be configured to be provided. Further, the control program may be provided by being stored on a computer connected to a network such as the Internet and downloaded via the network. Furthermore, the control program may be provided or distributed via a network such as the Internet.

  FIG. 9 exemplifies processing when determining the compression quantization table in the image processing apparatus 1 shown in FIG. When the pre-compression process is started, the pre-compression quantization table applied to the input image 25 (see FIG. 3) divided by the table switching control unit 21 into a plurality of regions (hereinafter abbreviated as a table in the description of the process). While switching between A, B, C, and D, the quantization unit 14 and the Huffman coding unit 15 perform quantization and coding processing (S1). The code amount counter 17 counts the code amount for each of the tables A, B, C, and D generated by the encoding process (S2). The main compression table determination unit 18 determines the relationship between the code amount of each table A, B, C, and D and the target code amount (S3).

  In step S3, when the target code amount is equal to or larger than the code amount corresponding to the table A and equal to or smaller than the code amount corresponding to the table B, the process shown in FIG. 10 is executed, and the target code amount is equal to or larger than the code amount corresponding to the table C. When the code amount is equal to or smaller than the code amount corresponding to the table D, the processing shown in FIG. 11 is executed. When the target code amount is larger than the code amount corresponding to the table D and smaller than the code amount corresponding to the table A, The process shown is executed.

In step S11 shown in FIG. 10, the following values are obtained.
Difference between code amount of table A and code amount of table B: L
Difference between Q value (Qa) of table A and Q value of table B: M
Increase amount of code amount per 1Q value: N (= L / M)
Difference between code amount of table A and target code amount: R
Increase amount of Q value corresponding to difference R: S (= R / N)
Q value for main compression: Qr (= Qa + S)

  Thereafter, the main compression dose table corresponding to Qr is extracted from the main compression table database 19 (S12) and used for quantization of the main compression.

  The processes in steps S11 and S12 in FIG. 10 are similarly performed in steps S21 and S22 in FIG. 11 and steps S31 and S32 in FIG.

  Below, the specific example of the process after step S3 of FIG. 9 is shown. For example, when the target code amount is 150 KB, the code amount corresponding to Table A is 250 KB, the code amount corresponding to Table B is 350 KB, the code amount corresponding to Table C is 100 KB, and the code amount corresponding to Table D is 180 KB Since the target code amount 150 KB is between the table C (100 KB) and the table D (180 KB), the relationship of “C ≦ target code amount ≦ D” is satisfied in step S 3.

In this case, the following values are obtained in step S21 of FIG.
Difference between code amount of table C and code amount of table D: L = 180−100 = 80
Difference between Q value (60) of table C and Q value (80) of table D: M = 80−60 = 20
Increase amount of code amount per 1Q value: N = 80/20 = 4
Difference between code amount of table C and target code amount: R = 150−100 = 50
Increase amount of Q value corresponding to difference R: S = 50/4 = 12.5
Q value for main compression: Qr = 60 + 12.5 = 72.5

  A main compression dose table corresponding to the Q value (Qr = 72.5) for main compression thus obtained is extracted from the main compression table database 19 (S22) and used for quantization of main compression.

  Although FIG. 2 shows a configuration in which the main compression quantization table is prepared in advance in the main compression table database 19, it is based on the main compression Q value (Qr) obtained as described above. In this case, a new compression quantization table may be created.

  According to the image processing apparatus 1, it is possible to perform preliminary compression for determining an optimum quantization table to be used at the time of main compression by only basic components necessary for normal compression processing. As a result, it is possible to realize highly accurate code amount control using basic components necessary for compression processing.

  The present invention is not limited to the above embodiment, and can be appropriately changed without departing from the gist of the invention. For example, FIG. 2 shows a form using the JPEG method, but the present invention should not be limited to this. Other than the JPEG, the code amount is not constant according to the characteristics of the input image. The present invention can also be applied to a compression method.

DESCRIPTION OF SYMBOLS 1 Image processing apparatus 2 Image division part 3 Quantization part 4 Encoding part 5 Table holding part 6 Table control part 7 Code amount detection part 8 Table determination part 11 Preprocessing part 12 Preprocessing memory 13 DCT part 14 Quantization part 15 Huffman coding unit 16 Code output control unit 17 Code amount counter 18 Main compression table determination unit 19 Main compression table database 20 Target code amount holding memory 21 Table switching control unit 22 Pre-compression table database 23 Main compression table holding memory 25 Input images

JP-A-5-63994 JP 2004-112345 A

Claims (8)

Image dividing means for dividing the input image into a plurality of regions;
Quantization means for quantizing the divided input image based on a quantization table;
Encoding means for encoding the quantized input image;
Table holding means for holding a plurality of preliminary compression quantization tables created in advance for different quality values;
Table control means for applying the preliminary compression quantization table different for each region at the time of preliminary compression and causing the quantization means to perform quantization of the input image;
Code amount detection means for detecting a code amount output by the encoding means;
Based on the code amount for each pre-compression quantization table detected during the pre-compression and the target code amount during the main compression, the code amount output by the encoding means during the main compression is the target code amount. Table determining means for determining the quantization table to be used at the time of the main compression,
An image processing apparatus comprising: The table holding means further holds a plurality of main compression quantization tables created for different quality values,
The table determining means extracts the main compression quantization table corresponding to the calculated quality value from the table holding means;
The image processing apparatus according to claim 1. The table determination means creates the quantization table used in the main compression based on the calculated quality value.
The image processing apparatus according to claim 1. The preliminary compression quantization table is created based on the correlation between the quality value and the code amount corresponding to the encoding method in the encoding means,
The image processing apparatus according to claim 1. The encoding method is JPEG.
The image processing apparatus according to claim 4.   The table holding means further holds a plurality of the preliminary compression quantization tables created in advance for each type of the input image,
  The table control means applies the preliminary compression quantization table corresponding to the type of the input image, and causes the quantization means to quantize the input image;
  The image processing apparatus according to claim 1. Dividing the input image into a plurality of regions;
Quantizing the divided input image based on a quantization table;
Encoding the quantized input image;
Applying a plurality of preliminary compression quantization tables created in advance for each different quality value for each region at the time of preliminary compression, and quantizing the input image;
Detecting a code amount output by the encoding;
Based on the code amount for each pre-compression quantization table detected during the pre-compression and the target code amount during the main compression, the code amount output by the encoding means during the main compression is the target code amount. Determining the quantization table to be used during the main compression so that
An image processing method including: In the computer that controls the image processing apparatus,
A process of dividing the input image into a plurality of areas;
A process of quantizing the divided input image based on a quantization table;
A process of encoding the quantized input image;
A process of quantizing the input image by applying a plurality of preliminary compression quantization tables created for each different quality value in advance for each region during preliminary compression,
A process of detecting a code amount output by the encoding;
Based on the code amount for each pre-compression quantization table detected during the pre-compression and the target code amount during the main compression, the code amount output by the encoding means during the main compression is the target code amount. Processing for determining the quantization table to be used at the time of the main compression,
An image processing program for executing

Images