JP5176228B2 - Encoder - Google Patents

Description

  The present invention relates to an encoder that compresses and encodes an input image into an output image.

  The encoder compresses and encodes an input image into an output image. The communication system can reduce the communication burden by communicating the output image after compression coding without communicating the input image before compression coding. The encoder of Patent Document 1 will be described with reference to FIGS.

  Input images D and E are input images on which compression encoding is performed. The image areas D1 and E1 are image areas that need to be displayed with higher definition than the image areas other than the image areas D1 and E1. The image areas D1 and E1 are, for example, face areas.

  In FIG. 7, three blocks D2, D3, and D4 distributed in the input image D are selected. A total of 100 target code amounts are allocated to the three blocks D2, D3, and D4. The block D3 is included in the image area D1. The blocks D2 and D4 are not included in the image area D1.

  A large amount of code is allocated to the block D3. The remaining code amount is allocated to the blocks D2 and D4. For example, a code amount of 50 is allocated for the block D3. For the blocks D2 and D4, 25 code amounts are allocated.

  In FIG. 7, a part of the input image can be displayed with high definition. The code amount can be reduced for the entire input image.

JP-A-8-340534

  In FIG. 8, three blocks E2, E3, E4 distributed in the input image E are selected. A total of 100 target code amounts are allocated to the three blocks E2, E3, and E4. Blocks E2, E3, and E4 are included in the image area E1.

  A large amount of code needs to be allocated for the blocks E2, E3, and E4. For example, it is necessary to allocate 33 code amounts for each of the blocks E2, E3, and E4. It is not easy to make the code amount allocated for each of the blocks E2, E3, and E4 equal to the code amount allocated for the block D3.

  In FIG. 8, it is not easy to reliably display a part of the input image with high definition. It is not easy to reliably reduce the code amount for the entire input image.

  Therefore, in view of the above problems, the present invention provides an encoder that compresses and encodes an input image into an output image, displays a part of the input image with higher definition, and encodes the entire input image more effectively. The purpose is to provide a technique for reducing the amount.

In order to solve the above-mentioned problem, the invention described in claim 1 is an encoder that compresses and encodes an input image into an output image, and a quantization step value setting unit that sets a first quantization step value for the input image. And scanning the input image in a first direction to classify the input image into a predetermined image region and a non-predetermined image region other than the predetermined image region, and the input image is separated into a second direction perpendicular to the first direction. By scanning in the direction, the input image is classified into a predetermined image region and a non-predetermined image region, and a predetermined image region and a non-predetermined image region separated by the scanning in the first direction. A first direction correction coefficient to which a value is assigned is set, and a second direction correction coefficient to which a different value is assigned to a predetermined image area and a non-predetermined image area that are separated by scanning in the second direction A correction coefficient setting unit that, by correcting the first quantization step value based on the calculated correction coefficient from the first direction correction coefficient and the second direction correction coefficient, the second quantization step value A quantization step value correction unit to be set, and a quantization unit that quantizes the input image based on the second quantization step value.

  According to a second aspect of the present invention, in the encoder according to the first aspect, the predetermined image area includes an image area that needs to be displayed with higher definition than the non-predetermined image area, and the predetermined image area includes the image area. The second quantization step value is smaller than the second quantization step value in the non-predetermined image region.

  According to a third aspect of the present invention, in the encoder according to the first aspect, the predetermined image area includes an image area that does not need to be displayed with higher definition than the non-predetermined image area, and the predetermined image area includes the image area. The second quantization step value is larger than the second quantization step value in the non-predetermined image region.

  According to a fourth aspect of the present invention, in the encoder according to any one of the first to third aspects, the quantization step value correction unit performs the first step for each of the predetermined image region and the non-predetermined image region. A multiplication correction coefficient setting unit for setting a multiplication correction coefficient to be multiplied by the quantization step value; and multiplying the first quantization step value by the multiplication correction coefficient for each of the predetermined image area and the non-predetermined image area And a correction coefficient multiplier for setting the second quantization step value.

  According to a fifth aspect of the present invention, in the encoder according to any one of the first to third aspects, the quantization step value correction unit performs the first step for each of the predetermined image region and the non-predetermined image region. An addition correction coefficient setting unit for setting an addition correction coefficient to be added to the quantization step value; and adding the addition correction coefficient to the first quantization step value for each of the predetermined image area and the non-predetermined image area And a correction coefficient adding unit for setting the second quantization step value.

  According to a sixth aspect of the present invention, in the encoder according to any one of the first to fifth aspects, the quantization step value correction unit is configured to determine the boundary region between the predetermined image region and the non-predetermined image region. A high-frequency component removing unit that removes the high-frequency component of the two-quantization step value;

  According to a seventh aspect of the present invention, in the encoder according to any one of the first to sixth aspects, the image area classification unit detects the coordinate position of the predetermined image area with respect to the input image. Means for notifying the quantization step value correction unit of the coordinate position of the predetermined image area.

  According to an eighth aspect of the present invention, in the encoder according to any one of the first to sixth aspects, the image area classification unit is configured to designate a coordinate position of the predetermined image area for the input image. Means for notifying the quantization step value correction unit of the coordinate position of the predetermined image area.

  The encoder of the present invention includes an image region classification unit and a quantization step value correction unit. The image region classification unit separates the predetermined image region and the non-predetermined image region other than the predetermined image region. The quantization step value correction unit corrects the quantization step value with different characteristics for each of the predetermined image region and the non-predetermined image region.

  When the predetermined image area needs to be displayed with higher definition than the non-predetermined image area, the quantization step value correction unit makes the quantization step value in the predetermined image area smaller than the quantization step value in the non-predetermined image area. In addition, the quantization step value is corrected.

  When the predetermined image area does not need to be displayed with higher definition than the non-predetermined image area, the quantization step value correction unit causes the quantization step value in the predetermined image area to be larger than the quantization step value in the non-predetermined image area. In addition, the quantization step value is corrected.

  The encoder of the present invention can display a part of the input image with higher definition. The encoder of the present invention can more effectively reduce the code amount for the entire input image.

{First embodiment}
[Encoder components]
Hereinafter, the first embodiment will be described with reference to the drawings. FIG. 1 is a diagram showing an encoder 1 according to the first embodiment. The encoder 1 compresses and encodes an input image including one or a plurality of face regions into an output image.

  The encoder 1 includes a frequency conversion unit 11, a quantization unit 12, an encoding unit 13, a quantization step value setting unit 14, a face area detection unit 15, and a quantization step value correction unit 16.

  The frequency conversion unit 11 performs frequency conversion on the input image. The quantization unit 12 performs quantization on the input image after frequency conversion. The encoding unit 13 performs encoding on the input image after frequency conversion and quantization.

  The quantization step value setting unit 14 sets a quantization step value. However, the quantization unit 12 does not perform quantization based on the quantization step value set by the quantization step value setting unit 14.

  The face area detection unit 15 detects a face area. For example, the face area detection unit 15 detects a face area by detecting a skin color image area. The horizontal direction scanning unit 15H detects the face area by scanning the input image in the horizontal direction. The vertical scanning unit 15V detects the face area by scanning the input image in the vertical direction.

  The quantization step value correction unit 16 corrects the quantization step value. The quantization unit 12 performs quantization based on the quantization step value corrected by the quantization step value correction unit 16.

  The quantization step value correction unit 16 includes a correction coefficient setting unit 161, a correction coefficient storage unit 162, a correction coefficient average unit 163, a high frequency component removal unit 164, and a correction coefficient multiplication unit 165.

  The correction coefficient setting unit 161 sets a correction coefficient. The correction coefficient setting unit 161 sets the correction coefficient in the face area to be smaller than the correction coefficient in the image area other than the face area.

  The correction coefficient storage unit 162 stores correction coefficients. The horizontal direction correction coefficient 162H is a correction coefficient set based on the horizontal direction scan. The vertical direction correction coefficient 162V is a correction coefficient set based on the vertical direction scan.

  The correction coefficient averaging unit 163 calculates an average value of the horizontal direction correction coefficient 162H and the vertical direction correction coefficient 162V. The high frequency component removal unit 164 removes the high frequency component of the average value of the correction coefficient in the boundary region between the face region and the image region other than the face region.

  The correction coefficient multiplication unit 165 receives the quantization step value before correction from the quantization step value setting unit 14. The correction coefficient multiplication unit 165 calculates the corrected quantization step value by multiplying the quantization step value before correction by the average value of the correction coefficient. The correction coefficient multiplication unit 165 outputs the corrected quantization step value to the quantization unit 12.

[Correction factor for quantization step value]
FIG. 2 is a diagram illustrating a correction coefficient for the quantization step value according to the first embodiment. The input image A is an input image on which compression encoding is executed. The image area A1 is a face area.

  Scan lines AH1, AH2, and AH3 indicate scanning directions by the horizontal scanning unit 15H. The scan line AH2 passes through the image area A1. The scan lines AH1 and AH3 do not penetrate the image area A1.

  Scan lines AV1, AV2, and AV3 indicate scan directions by the vertical direction scan unit 15V. The scan line AV2 passes through the image area A1. The scan lines AV1 and AV3 do not penetrate the image area A1.

  The horizontal solid line described below the input image A indicates the horizontal correction coefficient 162H stored in the correction coefficient storage unit 162 for the scan lines AH1, AH2, and AH3.

  In the scan lines AH1 and AH3, the horizontal direction correction coefficient 162H is 1.0 from the left end to the right end of the input image A. In the scan line AH2, the horizontal direction correction coefficient 162H adopts 1.0 in the portion that does not penetrate the image area A1, and the horizontal direction correction coefficient 162H is 0 smaller than 1.0 in the part that penetrates the image area A1. .8 is adopted.

  A solid line HH2 indicates that the horizontal direction correction coefficient 162H rapidly changes from 0.8 to 1.0 in the boundary region between the image region A1 and the image region other than the image region A1.

  The vertical solid line described on the right side of the input image A indicates the vertical correction coefficient 162V stored in the correction coefficient storage unit 162 for the scan lines AV1, AV2, and AV3.

  In the scan lines AV1 and AV3, the vertical direction correction coefficient 162V is 1.0 from the upper end to the lower end of the input image A. In the scan line AV2, the vertical direction correction coefficient 162V is 1.0 in a portion that does not pass through the image area A1, and the vertical direction correction coefficient 162V is 0 that is less than 1.0 in a part that passes through the image area A1. .8 is adopted.

  A solid line HV2 indicates that the vertical direction correction coefficient 162V rapidly changes from 0.8 to 1.0 in the boundary region between the image region A1 and the image region other than the image region A1.

  For the face region, the quantization step value is corrected so that the quantization step value becomes small. The encoder 1 can display the face area with higher definition. The encoder 1 can reduce the code amount more effectively for the entire input image.

  An alternate long and short dash line LH2 written on the lower side of the input image A indicates a horizontal direction correction coefficient in which the high frequency component is removed by the high frequency component removing unit 164 with respect to the scan line AH2.

  An alternate long and short dash line LH2 indicates that the horizontal direction correction coefficient gradually changes from 0.8 to 1.0 in the boundary region between the image region A1 and the image region other than the image region A1.

  A one-dot chain line LV2 described on the right side of the input image A indicates a vertical direction correction coefficient in which the high frequency component is removed by the high frequency component removing unit 164 with respect to the scan line AV2.

  An alternate long and short dash line LV2 indicates that the vertical direction correction coefficient gradually changes from 0.8 to 1.0 in the boundary region between the image region A1 and the image region other than the image region A1.

  The quantization step value is corrected so that the quantization step value gradually changes in the boundary region between the face region and the image region other than the face region. The encoder 1 can make a gradual transition from a high-definition image region to a non-high-definition image region in the boundary region between the face region and the image region other than the face region.

  In FIG. 2, both the horizontal direction scanning unit 15H and the vertical direction scanning unit 15V detect that the image area A1 is a face area. However, the horizontal direction scanning unit 15H and the vertical direction scanning unit 15V may make different determinations regarding the presence or absence of the face region, particularly in the boundary region between the face region and the image region other than the face region. That is, the horizontal direction correction coefficient 162H and the vertical direction correction coefficient 162V may adopt different values, particularly in the face area and the boundary area of the image area other than the face area.

  The face area detection unit 15 can detect the face area with higher accuracy even when both scanning units make different determinations regarding the presence or absence of the face area. The correction coefficient averaging unit 163 can calculate the correction coefficient as the correction coefficient in the boundary region by calculating the average value of both the correction coefficients even when the two correction coefficients adopt different values.

{Second Embodiment}
Next, a second embodiment will be described. FIG. 3 is a diagram illustrating the encoder 2 according to the second embodiment. The encoder 2 compresses and encodes an input image including the first image end into an output image. The first image edge is an image edge area added to the output image when the aspect ratio of the input resolution and the aspect ratio of the output resolution (or display resolution) are different.

  The encoder 2 includes a frequency conversion unit 21, a quantization unit 22, an encoding unit 23, a quantization step value setting unit 24, a first image end detection unit 25, and a quantization step value correction unit 26.

  The frequency conversion unit 21, the quantization unit 22, the encoding unit 23, the quantization step value setting unit 24, the first image end detection unit 25, and the quantization step value correction unit 26 are substantially the same as the corresponding components of the encoder 1. Execute the process.

  The first image edge detection unit 25 detects the first image edge. For example, the first image edge detection unit 25 detects the first image edge by detecting an image area with little motion or a black image area. The horizontal direction scanning unit 25H detects the first image edge by scanning the input image in the horizontal direction. The vertical scanning unit 25V detects the first image edge by scanning the input image in the vertical direction.

  The correction coefficient setting unit 261 sets a correction coefficient. The correction coefficient setting unit 261 sets the correction coefficient at the first image edge to be larger than 1.0, and sets the correction coefficient in the image area other than the first image edge in the same manner as in the first embodiment.

  The correction coefficient storage unit 262 stores the correction coefficient. The horizontal direction correction coefficient 262H is a correction coefficient set based on the horizontal direction scan. The vertical direction correction coefficient 262V is a correction coefficient set based on the vertical direction scan.

  FIG. 4 is a diagram illustrating a correction coefficient for the quantization step value according to the second embodiment. The input image B is an input image on which compression encoding is executed. Image regions B1 and B2 are the first image ends. In FIG. 4, the correction coefficient in the image area other than the image areas B1 and B2 is set to 1.0 for the sake of brevity. When an image area other than the image areas B1 and B2 includes a face area, correction coefficients in the image areas other than the image areas B1 and B2 may be set in the same manner as in the first embodiment.

  A scan line BH1 indicates a scanning direction by the horizontal scanning unit 25H. The scan line BH1 passes through the image areas B1 and B2.

  Scan lines BV1, BV2, and BV3 indicate the scanning direction by the vertical scanning unit 25V. The scan lines BV1 and BV3 pass through the image areas B1 and B2, respectively. The scan line BV2 does not penetrate the image areas B1 and B2.

  The horizontal solid line described below the input image B indicates the horizontal correction coefficient 262H stored in the correction coefficient storage unit 262 for the scan line BH1.

  In the scan line BH1, the horizontal correction coefficient 262H is 1.0 in a portion that does not pass through the image areas B1 and B2, and the horizontal correction coefficient 262H is 1 in a portion that passes through the image areas B1 and B2. Employs 1.2 greater than 0.

  The vertical solid line written on the right side of the input image B indicates the vertical correction coefficient 262V stored in the correction coefficient storage unit 262 for the scan lines BV1, BV2, and BV3.

  In the scan line BV2, the vertical correction coefficient 262V is 1.0 from the upper end to the lower end of the input image B. In the scan lines BV1 and BV3, the vertical correction coefficient 262V is 1.2, which is larger than 1.0, from the upper end to the lower end of the input image B.

  For the first screen edge, the quantization step value is corrected so that the quantization step value is increased. The encoder 2 can display the image area other than the first screen edge with higher definition. The encoder 2 can reduce the code amount more effectively for the entire input image.

  A solid line HH1 indicates a rapid change in the correction coefficient from 1.0 to 1.2 before the high-frequency component is removed. An alternate long and short dash line LH1 indicates a gradual change from 1.0 to 1.2 in the correction coefficient after the high frequency component is removed. The encoder 2 can make a gradual transition from an image area that is displayed with high definition to an image area that is not displayed with high definition in the boundary area between the first image end and the image area other than the first image end.

  In FIG. 4, both the horizontal direction scanning unit 25H and the vertical direction scanning unit 25V detect that the image regions B1 and B2 are the first image ends. However, the horizontal direction scanning unit 25H and the vertical direction scanning unit 25V may make different determinations regarding the presence or absence of the first image edge, particularly in the boundary region of the image area other than the first image edge and the first image edge. That is, different values may be adopted for the horizontal direction correction coefficient 262H and the vertical direction correction coefficient 262V, particularly in the boundary area between the first image edge and the image area other than the first image edge.

  The first image edge detection unit 25 can detect the first image edge more accurately even when both scanning units make different determinations regarding the presence or absence of the first image edge. The correction coefficient averaging unit 263 can calculate a correction coefficient as a correction coefficient in the boundary region by calculating an average value of both correction coefficients even when both correction coefficients adopt different values.

{Third embodiment}
Next, a third embodiment will be described. FIG. 5 is a diagram illustrating the encoder 3 according to the third embodiment. The encoder 3 compresses and encodes an input image including the second image end into an output image. The second image edge is an image edge area added to the output image when the aspect ratio of the input resolution and the aspect ratio of the output resolution (or display resolution) are different.

  The encoder 3 includes a frequency conversion unit 31, a quantization unit 32, an encoding unit 33, a quantization step value setting unit 34, a second image end detection unit 35, and a quantization step value correction unit 36.

  The frequency conversion unit 31, the quantization unit 32, the encoding unit 33, the quantization step value setting unit 34, the second image edge detection unit 35, and the quantization step value correction unit 36 are substantially the same as the corresponding components of the encoder 1. Execute the process.

  The second image edge detection unit 35 detects the second image edge. However, the second image edge detection unit 35 does not detect the second image edge by scanning the input image. Rather, the second image edge detection unit 35 detects the second image edge by designating the screen size and the display size from the outside of the encoder 3.

  The correction coefficient setting unit 361 sets a correction coefficient. The correction coefficient setting unit 361 sets the correction coefficient at the second image edge to be larger than 1.0, and sets the correction coefficient in the image area other than the second image edge in the same manner as in the first embodiment.

  The correction coefficient storage unit 362 stores correction coefficients. Here, the second image edge detection unit 35 does not scan the input image. Therefore, the correction coefficient storage unit 362 does not store the horizontal direction correction coefficient and the vertical direction correction coefficient for each block of the input image. Rather, the correction coefficient storage unit 362 stores a single correction coefficient for each block of the input image.

  The second image edge detection unit 35 does not scan the input image. The quantization step value correction unit 36 does not need to calculate the average value of the horizontal direction correction coefficient and the vertical direction correction coefficient, and therefore does not include the correction coefficient average unit.

  The second image end detection unit 35 designates the screen size and the display size from the outside of the encoder 3. The quantization step correction unit 36 does not need to make a gradual transition from the second image end to the image region other than the second image end, and therefore does not include the high frequency component removal unit.

  FIG. 6 is a diagram illustrating the correction coefficient of the quantization step value according to the third embodiment. The input image C is an input image on which compression encoding is executed. Image regions C1 and C2 are second image edges. Lines CH1, CH2, CH3, and CV1 are not actually scanned lines, but are lines for clearly indicating the correction coefficient. In FIG. 6, the correction coefficient in the image area other than the image areas C1 and C2 is set to 1.0 for the sake of brevity. When the image area other than the image areas C1 and C2 includes a face area, the correction coefficient in the image area other than the image areas C1 and C2 may be set in the same manner as in the first embodiment.

  Lines CH1 and CH3 pass through the image areas C1 and C2, respectively. The line CH2 does not penetrate the image areas C1 and C2. The line CV1 passes through the image areas C1 and C2.

  The solid line in the horizontal direction described below the input image C indicates the correction coefficient stored in the correction coefficient storage unit 362 for the lines CH1, CH2, and CH3.

  In the line CH2, a correction coefficient of 1.0 is adopted from the left end to the right end of the input image C. In the lines CH1 and CH3, the correction coefficient 1.2 larger than 1.0 is adopted from the left end to the right end of the input image C.

  A vertical solid line described on the right side of the input image C indicates a correction coefficient stored in the correction coefficient storage unit 362 for the line CV1.

  In the line CV1, a correction coefficient of 1.0 is adopted in a portion not penetrating the image areas C1 and C2, and a correction coefficient of 1.2 larger than 1.0 is adopted in a portion penetrating the image areas C1 and C2. adopt.

  For the second screen edge, the quantization step value is corrected so that the quantization step value becomes large. The encoder 3 can display the image area other than the second screen end with higher definition. The encoder 3 can reduce the code amount more effectively for the entire input image.

{Modification}
In the first to third embodiments, the quantization step value setting units 14, 24, and 34 may set the same quantization step value for each block of the input image before correction. In the first and second embodiments, the high frequency component removing units 164 and 264 may remove the high frequency component from the correction step or the quantization step value multiplied by the correction factor.

  In the first to third embodiments, the quantization step value setting units 14, 24, and 34 may set different quantization step values for each block of the input image before correction. In the first and second embodiments, the high frequency component removal units 164 and 264 may remove high frequency components from the quantization step value multiplied by the correction coefficient.

  In the first and second embodiments, the correction coefficient averaging units 163 and 263 calculate the average value of the horizontal direction correction coefficient and the vertical direction correction coefficient. Here, the multiplication unit may multiply the horizontal direction correction coefficient and the vertical direction correction coefficient. Alternatively, the selection unit may select any one of the horizontal direction correction coefficient and the vertical direction correction coefficient.

  In the first embodiment, consider a case where the horizontal direction correction coefficient is 0.8 and the vertical direction correction coefficient is 1.0 for the same image region. The multiplication unit multiplies 0.8 and 1.0 and outputs 0.8 as a correction coefficient to high frequency component removal unit 164. The selection unit selects 0.8 out of 0.8 and 1.0, and outputs 0.8 as a correction coefficient to the high frequency component removal unit 164. Similar to the correction coefficient averaging unit 163, the multiplication unit or the selection unit can calculate a correction coefficient as a correction coefficient in the boundary region in the boundary region between the face region and the image region other than the face region.

  In the second embodiment, consider a case where the horizontal correction coefficient is 1.2 and the vertical correction coefficient is 1.0 for the same image area. The multiplication unit multiplies 1.2 and 1.0, and outputs 1.2 as a correction coefficient to the high frequency component removal unit 264. The selection unit selects 1.2 out of 1.2 and 1.0, and outputs 1.2 as a correction coefficient to the high frequency component removal unit 264. Similar to the correction coefficient averaging unit 263, the multiplication unit or the selection unit can calculate a correction coefficient as a correction coefficient in the boundary region in the boundary region of the image region other than the first image end and the first image end.

  In the first to third embodiments, the correction coefficient multipliers 165, 265, and 363 calculate the corrected quantization step value by multiplying the quantization step value before correction by the correction coefficient. Here, the correction coefficient adding unit may calculate the corrected quantization step value by adding the correction coefficient to the uncorrected quantization step value. Alternatively, the correction coefficient subtraction unit may calculate the corrected quantization step value by subtracting the correction coefficient from the pre-correction quantization step value.

  The correction coefficient adding unit adds a positive correction coefficient to the quantization step value before correction for an image area that does not need to be displayed in high definition, and corrects the image area that needs to be displayed in high definition. For example, the previous quantization step value may be output as it is. The correction coefficient subtraction unit subtracts a positive correction coefficient from the quantization step value before correction for an image area that needs to be displayed in high definition, and corrects the image area that does not need to be displayed in high definition. For example, the previous quantization step value may be output as it is.

  In the first to third embodiments, the encoders 1, 2, and 3 may be arranged in the transcoder. The first stage decoder decodes the input image into a decoded image. The second stage encoders 1, 2, and 3 encode the decoded image into an output image. Decoders and encoders 1, 2, and 3 may adopt different or the same encoding method.

  The quantization step value setting units 14, 24, and 34 may set the quantization step values in the encoders 1, 2, and 3 based on the quantization step values in the decoder. The face area detection unit 15, the first image end detection unit 25, and the second image end detection unit 35 may specify the coordinate positions of the face area, the first image end, and the second image end from the decoder.

It is a figure which shows the encoder in 1st Embodiment. It is a figure which shows the correction coefficient of the quantization step value in 1st Embodiment. It is a figure which shows the encoder in 2nd Embodiment. It is a figure which shows the correction coefficient of the quantization step value in 2nd Embodiment. It is a figure which shows the encoder in 3rd Embodiment. It is a figure which shows the correction coefficient of the quantization step value in 3rd Embodiment. It is a figure which shows the allocation method of the code amount in a prior art. It is a figure which shows the allocation method of the code amount in a prior art.

Explanation of symbols

1, 2, 3 Encoder 11, 21, 31 Frequency conversion unit 12, 22, 32 Quantization unit 13, 23, 33 Encoding unit 14, 24, 34 Quantization step value setting unit 15 Face region detection unit 25 First image Edge detection unit 35 Second image edge detection unit 15H, 25H Horizontal scanning unit 15V, 25V Vertical scanning unit 16, 26, 36 Quantization step value correction unit 161, 261, 361 Correction coefficient setting unit 162, 262, 362 Correction Coefficient storage units 162H, 262H Horizontal direction correction coefficients 162V, 262V Vertical direction correction coefficients 163, 263 Correction coefficient averaging units 164, 264 High frequency component removal units 165, 265, 363 Correction coefficient multiplication units

Claims (8)

An encoder that compresses and encodes an input image into an output image,
A quantization step value setting unit for setting a first quantization step value for the input image;
By scanning the input image in the first direction, the input image is separated into a predetermined image region and a non-predetermined image region other than the predetermined image region, and the input image is moved in a second direction perpendicular to the first direction. An image area classification unit that separates the input image into a predetermined image area and a non-predetermined image area by scanning ;
A first direction correction coefficient assigned with different values is set for the predetermined image region and non-predetermined image region sorted by the scan in the first direction, and the predetermined image region and non-predetermined by the second direction scan are set. A correction coefficient setting unit that sets a second direction correction coefficient that is assigned a different value to the image area;
Quantization step value correction for setting a second quantization step value by correcting the first quantization step value based on a correction coefficient calculated from the first direction correction coefficient and the second direction correction coefficient And
A quantization unit that quantizes the input image based on the second quantization step value;
An encoder comprising: The encoder according to claim 1, wherein
The predetermined image area is
An image area that needs to be displayed with higher definition than the non-predetermined image area;
Including
The encoder according to claim 1, wherein the second quantization step value in the predetermined image region is smaller than the second quantization step value in the non-predetermined image region. The encoder according to claim 1, wherein
The predetermined image area is
An image area that does not need to be displayed with higher definition than the non-predetermined image area;
Including
The encoder according to claim 1, wherein the second quantization step value in the predetermined image region is larger than the second quantization step value in the non-predetermined image region. The encoder according to any one of claims 1 to 3,
The quantization step value correction unit
A multiplication correction coefficient setting unit for setting a multiplication correction coefficient to be multiplied by the first quantization step value for each of the predetermined image area and the non-predetermined image area;
A correction coefficient multiplier for setting the second quantization step value by multiplying the first quantization step value by the multiplication correction coefficient for each of the predetermined image area and the non-predetermined image area;
The encoder characterized by including. The encoder according to any one of claims 1 to 3,
The quantization step value correction unit
An addition correction coefficient setting unit for setting an addition correction coefficient to be added to the first quantization step value for each of the predetermined image area and the non-predetermined image area;
A correction coefficient addition unit that sets the second quantization step value by adding the addition correction coefficient to the first quantization step value for each of the predetermined image region and the non-predetermined image region;
The encoder characterized by including. The encoder according to any one of claims 1 to 5,
The quantization step value correction unit
A high frequency component removing unit that removes a high frequency component of the second quantization step value for a boundary region between the predetermined image region and the non-predetermined image region
The encoder characterized by including. The encoder according to any one of claims 1 to 6,
The image area classification unit includes:
Means for notifying the quantization step value correction unit of the coordinate position of the predetermined image area by detecting the coordinate position of the predetermined image area for the input image;
The encoder characterized by including. The encoder according to any one of claims 1 to 6,
The image area classification unit includes:
Means for notifying the quantization step value correction unit of the coordinate position of the predetermined image area by designating the coordinate position of the predetermined image area for the input image;
The encoder characterized by including.

Images