JP4400160B2 - Image processing device - Google Patents

Description

The present invention relates to a technique for enhancing an edge of a digital image to obtain an image with good sharpness.

  In general, when displaying or printing a digital image obtained by photographing with a digital camera or the like, contour enhancement processing is performed on the original image in order to obtain an image with good sharpness.

  Conventionally, in the edge enhancement processing of a digital image, a high frequency component of an image is obtained from a difference between an image signal of the original image and an unsharp signal obtained by weighted averaging of the image signal of the original image, and the high frequency component of the image is obtained as the original image. An image with an enhanced outline is obtained by adding to the image.

  At this time, for the high frequency component of the image obtained by the difference from the unsharp signal of the original image, in order to remove the noise contained in the high frequency component of the image, the absolute value of this image high frequency component Is smaller than a predetermined threshold value, it is set to “0” or a value close to “0” and added to the original image.

Such an image edge enhancement method is called an unsharp mask method, and is adopted as an image processing function in a photographic image printing apparatus or the like (see, for example, Patent Document 1).
JP-A-10-243238

  However, in the conventional contour enhancement processing, the high-frequency component H of the image is a difference value between the target pixel and an unsharp signal by a weighted average such as a Gaussian type in an n × n pixel region centered on the target pixel. Therefore, the characteristic is sensitive to isolated point noise of an image which is a kind of high frequency component, and the obtained image high frequency component is added to the original image as an image contour component or “0” as a noise component. Alternatively, it is difficult to distinguish whether the value is close to “0” and added to the original image, and it is very difficult to set the threshold value to an appropriate value.

  That is, if the threshold value for the high frequency component H of the image is set large, it is possible to reliably prevent the noise component from being added and emphasized, but the contour component is also included in the removal target, and the image having poor sharpness. End up.

  Conversely, if the threshold value for the high frequency component H of the image is set small, not only the contour component but also the noise component will be emphasized, resulting in an image with noticeable noise.

The present invention has been made in view of such a problem, and an object of the present invention is to provide a technique capable of obtaining a contour-enhanced image in which a noise component of an original image is appropriately suppressed.

The present invention
High-frequency component extraction means for extracting high-frequency components of the image;
Contour enhancement component generating means for generating a contour enhancement component from the high frequency component;
Addition means for acquiring a contour-enhanced image in which the contour on the image is enhanced by adding the contour-enhanced component to the image; and
The outline emphasis component generation means includes
Coring means for outputting, as the contour emphasis component, a component consisting only of a target pixel equal to or higher than a predetermined threshold among the target pixels constituting the high-frequency component;
Middle region contour component detecting means for detecting a contour component of the middle region of the image;
High frequency contour component detecting means for detecting a high frequency contour component of the image;
For the target pixel, calculation means for calculating a contour component value based on the contour component of the middle region and the contour component of the high region,
When the contour component value is larger than a predetermined value, the threshold value is set to a first value, and when the contour component value is smaller than the predetermined value, the threshold value is set to a second value higher than the first value. Threshold setting means for setting to

  According to this, according to the size of the middle and high frequency contour component of the image data, the image high frequency component corresponding to the image noise is appropriately removed, and the image high frequency component corresponding to the image contour is extracted as the contour enhancement component. As a result, a contour-enhanced image with good sharpness can be acquired.

  According to the image processing apparatus according to the present invention, the image high frequency component corresponding to the image noise is appropriately removed according to the size of the middle and high frequency contour component of the image data, and the image high frequency component corresponding to the image contour is removed. It can be taken out as a contour enhancement component, and a contour-enhanced image with good sharpness can be acquired.

Therefore, according to the present invention, it is possible to obtain an edge enhanced image in which the noise component of the original image is appropriately suppressed .

  Embodiments of the present invention will be described below with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram showing a configuration of a digital camera having an edge enhancement function according to an embodiment of an image processing apparatus of the present invention.

  This digital camera includes an imaging unit 10 using a CCD (charge coupled device) as an imaging device. An image pickup signal of the image obtained by the image pickup unit 10 is converted into a digital image signal via the A / D conversion unit 11 and input to the image interpolation circuit 12 as Bayer data.

  The image interpolation circuit 12 converts the input Bayer data into RGB data for each pixel. The image data converted into the RGB data is input to the matrix circuit 13 to be inputted with a luminance signal (Y) and a color difference signal ( RY) (BY).

  The luminance signal (Y) and color difference signals (RY) (BY) obtained by the mat risk circuit 13 are input to the image processing unit 14.

  The image processing unit 14 includes an edge enhancement processing unit 15, and is based on the luminance signal Y input from the matrix circuit 13 in accordance with a control signal corresponding to an image processing program stored in advance from the control unit (CPU) 16. An edge-enhanced image is generated, and various image processing such as density adjustment processing and hue adjustment processing based on the color difference signals (RY) (BY) is performed.

  The control unit (CPU) 16 is also WB (white balance), AE (auto exposure), WB based on the luminance signal (Y) and color difference signals (RY) (BY) input from the matrix circuit 13. Each control signal of AF (auto focus) is generated and output to the imaging unit 10.

  The image data subjected to various image processes in the image processing unit 14 is recorded in the image memory 17, written in the display buffer 18, and displayed on the display unit 19.

  The image data recorded in the image memory 17 is output to a printer (not shown) and printed, or output to an external monitor and displayed on a monitor.

  FIG. 2 is a block diagram showing a configuration of the contour enhancement processing unit 15 of the first embodiment provided in the image processing unit 14 of the digital camera.

  The contour emphasis processing unit 15 of the first embodiment obtains a contour emphasis component Se by setting a threshold T to the image high-frequency component Sh and a high-frequency component extraction circuit 21 that extracts the image high-frequency component Sh from the original image Si. A coring circuit 22; an adder 23 for adding a contour emphasis component Se to the original image Si to obtain a contour emphasis image So; and a contour detection circuit (middle region) for detecting a middle band component E1 among the contour components of the original image Si. ) 24, a contour detection circuit (high frequency) 25 for detecting the high frequency component E2 of the contour components of the original image Si, a coefficient unit 26 for multiplying the mid frequency contour component E1 by a predetermined coefficient K1, and a high frequency contour component E2. A coefficient multiplier 27 that multiplies a predetermined coefficient K2, an adder 28 that adds the output of the coefficient multiplier 26 and the output of the coefficient multiplier 27 to obtain a contour component E, and a threshold for the coring circuit 22 in accordance with a change in the contour component E Threshold for variably setting T It is configured to include a control circuit 29.

  FIG. 3 is a diagram showing an example of a filter coefficient used in the high frequency component extraction circuit 21 of the contour enhancement processing unit 15, and FIG. 3A directly shows the image high frequency by multiplying the original image Si. FIG. 5B is a diagram showing a Laplacian coefficient 21a for obtaining the component Sh, and FIG. 5B shows an image obtained by multiplying the original image Si when subtracting the image low-frequency component SI from the original image Si to obtain the image high-frequency component Sh. It is a figure which shows the Laplacian coefficient 21b for obtaining low-pass component SI.

  The high-frequency component extraction circuit 21 obtains the image high-frequency component Sh by multiplying the target pixel and the adjacent four pixels by the Laplacian coefficient 21a shown in FIG. 3A for each pixel of the original image Si.

  FIG. 4 is a block diagram showing a configuration of another embodiment of the high frequency component extraction circuit 21 in the contour enhancement processing unit 15.

  The high-frequency component extraction circuit 21 of this other embodiment includes a weighted average circuit 211 and a subtractor 212. The original image Si is input to the weighted average circuit 211 to acquire the image low-frequency component SI, and then the subtractor 212. The image high-frequency component Sh is obtained by subtracting the image low-frequency component SI from the original image Si.

  In this case, the weighted average circuit 211 multiplies the original image Si to obtain the image low-frequency component SI by setting the weighted average coefficient 21b having the value shown in FIG. Can output the image high-frequency component Sh having the same value as the image high-frequency component Sh obtained by simply multiplying the original image Si by the Laplacian coefficient 21a shown in FIG.

  The image high-frequency component Sh thus extracted by the high-frequency component extraction circuit 21 is obtained by approximating the signal component equal to or less than “0” or “0” in the coring circuit 22 with the signal component equal to or lower than the threshold T set by the threshold control circuit 29 By using the value, the edge enhancement component Se in which the noise component is suppressed is obtained. Then, the edge emphasis component So obtained from the coring circuit 22 is added to the original image Si by the adder 23 to obtain the edge emphasis image So.

  On the other hand, the original image Si is input to the first contour detection circuit (middle region) 24 and the second contour detection circuit (high region) 25, and the first contour detection circuit (middle region) 24 outputs the original image Si. Among the contour components, the middle band component E1 is detected. The second contour detection circuit (high frequency) 25 detects a high frequency component E2 among the contour components of the original image Si.

  The mid-range contour component E1 and the high-frequency contour component E2 obtained by the first contour detection circuit (middle region) 24 and the second contour detection circuit (high region) 25 are respectively converted into coefficient units 26 and 27. The balance between the mid-frequency component and the high-frequency component is adjusted by multiplying by the coefficients K1 and K2, and added to the contour component E by the adder 28 and then input to the threshold value control circuit 29.

  FIG. 5 is a block diagram showing a configuration of the contour detection circuit 24 (25) in the contour enhancement processing unit 15. The contour detection circuit 24 (25) has a common configuration for both the first contour detection circuit (middle region) 24 and the second contour detection circuit (high region) 25.

  The contour detection circuit 24 (25) includes a contour detection circuit (horizontal) 30 for obtaining a horizontal contour component Eh and its absolute value circuit 32, a contour detection circuit (vertical) 31 for obtaining a vertical contour component Ev, and The absolute value conversion circuit 33 is provided, and the horizontal contour component Eh and the vertical contour component Ev are added by an adder 34 to obtain a contour component E (middle region contour component E1 or high region contour component E2).

  FIG. 6 is a diagram showing an operator coefficient for detecting a contour component used in the first contour detection circuit (middle region) 24 of the contour enhancement processing unit 15, and FIG. 6A shows a horizontal contour component (middle region). The figure which shows the coefficient 24h for a detection, The figure (B) is a figure which shows the coefficient 24v for a vertical contour component (middle area) detection.

  That is, in the first contour detection circuit (middle region) 24, the contour detection circuit (horizontal) 30 shows the target pixel of the original image Si and the adjacent 3 × 3 pixels as shown in FIG. The horizontal contour component Eh1 of the middle region is obtained by multiplying the horizontal contour component (middle region) detection coefficient 24h. Similarly, in the contour detection circuit (vertical) 31, the target pixel of the original image Si and 3 × 3 adjacent pixels are obtained. Is multiplied by the vertical contour component (middle region) detection coefficient 24v shown in FIG. 6B to obtain the middle region vertical contour component Ev1. Then, the horizontal contour component (middle region) Eh1 and the vertical contour component (middle region) Ev1 absolute valued by the absolute value circuits 32 and 33 are added by the adder 34 to obtain the middle region contour component E1.

  FIG. 7 is a diagram showing an operator coefficient for detecting a contour component used in the second contour detection circuit (high frequency) 25 of the contour enhancement processing unit 15, and FIG. 7A shows a horizontal contour component (high frequency). The figure which shows the coefficient 25h for a detection, The figure (B) is a figure which shows the coefficient 25v for a vertical contour component (high region) detection.

  That is, in the second contour detection circuit (high frequency) 25, the contour detection circuit (horizontal) 30 shows the target pixel of the original image Si and the adjacent 3 × 3 pixels as shown in FIG. The horizontal contour component (high frequency) detection coefficient 25h is multiplied to obtain a high frequency horizontal contour component Eh2. Similarly, in the contour detection circuit (vertical) 31, the target pixel of the original image Si and 3 × 3 adjacent pixels are obtained. Is multiplied by the vertical contour component (high frequency) detection coefficient 25v shown in FIG. 7B to obtain the vertical contour component Ev2 of the high frequency. Then, the horizontal contour component (high region) Eh2 and the vertical contour component (high region) Ev2 absolute valued by the absolute value circuits 32 and 33 are added by the adder 34 to obtain the high region contour component E2.

  The final contour component E1 (E2) is preferably the root mean square of the horizontal contour component Eh1 (Eh2) and the vertical contour component Ev1 (Ev2). Similar effects can be obtained.

  Thus, the adder 28 adds the mid-range contour component E1 obtained by the first contour detection circuit (mid-range) 24 and the high-frequency contour component E2 obtained by the second contour detection circuit (high range) 25. Then, it is input to the threshold value control circuit 29 as the contour component E for the target pixel of the original image Si.

  When the input contour component E of the target pixel is smaller than the set value, the threshold control circuit 29 determines that the image component of the target pixel is noise, and sets the threshold T for the coring circuit 22 to be higher. On the contrary, when the contour component E of the target pixel to be input is equal to or greater than the set value, the image component of the target pixel is determined to be a contour component, and the threshold T for the coring circuit 22 is set to be low.

  Next, a specific operation in the contour emphasis processing unit 15 of the first embodiment provided in the image processing unit 14 of the digital camera having the above configuration will be described.

  FIG. 8 is a diagram illustrating a specific example of image data input as the original image Si to the contour enhancement processing unit 15, and FIG. 8A is a diagram illustrating the image data Pa when there is isolated point noise in the center pixel. FIG. 7B is a diagram showing image data Pb when the middle vertical contour component Ev1 is high in the central pixel, and FIG. 10C shows image data when the high vertical contour component Ev2 is high in the central pixel. It is a figure which shows Pc.

  FIG. 9 shows that when the image data Pa, Pb, and Pc shown in FIG. 8 are input as the original image Si to the contour enhancement processing unit 15, the image high-frequency component Sh and the mid-region contour component having the respective central pixels as target pixels. It is a figure which contrasts and shows E1, the high region outline component E2, and the outline component E.

  Here, the coefficients K1 and K2 in the coefficient multipliers 26 and 27 are set to K1 = K2 = 1, and the outputs of the contour components E1 and E2 based on the respective contour component detection coefficients 24h, 24v, 25h, and 25v are The value is normalized by the sum of the absolute values of the coefficients.

  First, as shown in FIG. 8A, when image data Pa having isolated point noise at the center pixel is input as the original image Si to the contour enhancement processing unit 15 and the center pixel is the target pixel, When the image high frequency component Sh is obtained by multiplying the Laplacian coefficient 21a shown in FIG. 3A in the component extraction circuit 21, as shown in correspondence with the image Pa in FIG. 9, the image high frequency component Sh = 50. Is obtained.

  Further, in each of the horizontal and vertical contour detection circuits 30 and 31 in the first contour detection circuit (middle region) 24, the horizontal contour component shown in FIG. 6A is applied to the image data Pa which is the original image Si. By multiplying the (middle region) detection coefficient 24h and the vertical contour component (middle region) detection coefficient 24v shown in FIG. 6B, the horizontal and vertical contour components Eh1, Ev1 of the middle region are obtained, and these are obtained as absolute values. When the mid-region contour component E1 is obtained by adding and adding by the adder 34, the mid-region contour component E1 = 0 is obtained as shown in correspondence with the image Pa in FIG.

  Further, in each of the horizontal and vertical contour detection circuits 30 and 31 in the second contour detection circuit (high frequency) 25, the horizontal contour component shown in FIG. 7A is applied to the image data Pa which is the original image Si. By multiplying the high frequency detection coefficient 25h and the vertical contour component (high frequency) detection coefficient 25v shown in FIG. 7B, high frequency horizontal and vertical contour components Eh2 and Ev2 are obtained, which are absolute values. When the high frequency contour component E2 is obtained by adding the values in the adder 34, the high frequency contour component E2 = 33.3 is obtained as shown in correspondence with the image Pa in FIG.

  Then, the adder 28 adds the mid-range contour component E1 = 0 and the high-frequency contour component E2 = 33.3 to obtain the contour component E = 33.3 of the image data Pa, and a threshold value control circuit. 29.

  Next, as shown in FIG. 8B, image data Pb in the case where the middle vertical contour component Ev1 is high in the central pixel is input as the original image Si to the contour enhancement processing unit 15, and the central pixel is the target pixel. When the image high frequency component Sh is obtained by multiplying the Laplacian coefficient 21a in the same manner as described above, the high frequency component extraction circuit 21 obtains the image high frequency component Sh as shown in correspondence with the image Pb in FIG. = 12.5 is obtained.

  Further, in the first contour detection circuit (middle region) 24 (30 to 34), the horizontal contour component (middle region) detection coefficient 24h and the vertical contour component are similarly applied to the image data Pb as the original image Si. (Middle region) The horizontal and vertical contour components Eh1 and Ev1 of the middle region are obtained by multiplying by the detection coefficient 24v, and when the middle region contour component E1 is obtained by adding them to absolute values, the image Pb in FIG. As shown by corresponding to, a mid-range contour component E1 = 50 is obtained.

  In the second contour detection circuit (high region) 25 (30 to 34), the horizontal contour component (high region) detection coefficient 25h and the vertical contour component are similarly applied to the image data Pb as the original image Si. When the high frequency horizontal and vertical contour components Eh2 and Ev2 are obtained by multiplying by the (high frequency) detection coefficient 25v, and the high frequency contour component E2 is obtained by adding them to absolute values, the image Pb in FIG. As shown in FIG. 5, a high frequency contour component E2 = 25 is obtained.

  Then, the middle region contour component E1 = 50 and the high region contour component E2 = 25 are added to obtain the contour component E = 75 of the image data Pb and input to the threshold control circuit 29.

  Further, as shown in FIG. 8C, image data Pc when the high-frequency vertical contour component Ev2 is high in the central pixel is input as the original image Si to the contour enhancement processing unit 15, and the central pixel is the target pixel. In some cases, when the image high-frequency component Sh is obtained by multiplying the Laplacian coefficient 21 in the same manner as described above in the high-frequency component extraction circuit 21, as shown in correspondence with the image Pc in FIG. 9, the image high-frequency component Sh = 25 is obtained.

  In the first contour detection circuit (middle region) 24 (30 to 34), the horizontal contour component (middle region) detection coefficient 24h and the vertical contour component are similarly applied to the image data Pc as the original image Si. (Middle region) The horizontal and vertical contour components Eh1 and Ev1 of the middle region are obtained by multiplying by the detection coefficient 24v, and when the middle region contour component E1 is obtained by adding them to absolute values, the image Pc in FIG. 9 is obtained. As shown by corresponding to, a mid-range contour component E1 = 0 is obtained.

  In the second contour detection circuit (high region) 25 (30 to 34), the horizontal contour component (high region) detection coefficient 25h and the vertical contour component are similarly applied to the image data Pc as the original image Si. When the high frequency horizontal and vertical contour components Eh2 and Ev2 are obtained by multiplying by the (high frequency) detection coefficient 25v, and the high frequency contour component E2 is obtained by adding them to absolute values, the image Pc in FIG. 9 is obtained. As shown in FIG. 5, a high frequency contour component E2 = 50 is obtained.

  Then, the contour component E = 50 of the image data Pc is obtained by adding the middle region contour component E1 = 0 and the high region contour component E2 = 50, and is input to the threshold value control circuit 29.

  That is, the image high-frequency component Sh for the target pixel of each of the image data Pa, Pb, and Pc is large in the order of Pa> Pc> Pb, whereas the contour component E = E1 + E2 is Pb> Pc> Pa. In this order, the target pixel of the image data Pa can be properly determined to be isolated point noise with a small contour component E even if the image high-frequency component Sh is large. The target pixels of the image data Pb and Pc can be properly determined that the contour component E is large and the image contour even if the image high-frequency component Sh is small.

  In the threshold control circuit 29, when the contour component E obtained by adding the middle region contour component E1 and the high region contour component E2 in the target image of the original image Si is small, for example, less than “50”, coring is performed. The extraction threshold value T of the image high-frequency component Sh as the contour enhancement component Se for the circuit 22 is set to a high value, for example, “70” or more, and conversely, the mid-frequency contour component E1 and the high-frequency contour in the target image of the original image Si When the contour component E obtained by the addition of the component E2 is as large as, for example, “50” or more, the extraction threshold T as the contour enhancement component Se of the image high-frequency component Sh for the coring circuit 22 is set to, for example, “10”. Setting control is as low as above.

  Thereby, in the coring circuit 22, the image high frequency component Sh of the image data Pa in which the target pixel is isolated point noise as shown in FIG. 8A is removed even if the high frequency component Sh is large. The image high frequency component Sh of the image data Pb and Pc when the contour component E of the target pixel is large as shown in FIGS. 8B and 8C is not extracted as the contour emphasizing component Se. Even if the component Sh is small, it is extracted as the contour emphasizing component Se without being removed. Therefore, it is possible to obtain a contour-enhanced image So in which only the contour-enhanced component Se of the correct image contour from which the noise component has been appropriately removed is added to the original image Si.

  Therefore, according to the digital camera having the contour enhancement function of the first embodiment having the above-described configuration, the image high-frequency component Sh extracted from the original image Si is extracted as the contour enhancement component Se by the coring circuit 22 in which the threshold T is set. In addition, when the contour-enhanced image So is obtained by adding to the original image Si, a middle-region contour component E1 obtained by adding the horizontal contour component Eh1 and the vertical contour component Ev1 in the middle region of the original image Si, and a high-frequency region A contour component E (= E1 + E2) composed of a high-frequency contour component E2 obtained by adding the horizontal contour component Eh2 and the vertical contour component Ev2 is obtained, and when the middle-high frequency contour component E of the original image Si is large, the coring circuit 22 The threshold value T is controlled to be low so that the contour enhancement component Se is extracted even if the image high-frequency component Sh is small. Since the threshold value T of 22 is controlled to be high so that the image high-frequency component Sh is not extracted as the contour emphasizing component Se, it is possible to prevent image noise from being erroneously determined as an image contour and to be extracted, and to reduce noise. A contour-enhanced image So can be obtained.

  In the threshold value control circuit 29 in the contour enhancement processing unit 15 of the first embodiment, the threshold value T set in the coring circuit 22 is set according to the size of the contour component E (= E1 + E2) of the original image Si, for example. The setting control is performed in two stages of low / high. However, the setting control may be performed so as to change continuously according to the size of the contour component E (= E1 + E2).

  In the edge emphasis processing unit 15 of the first embodiment, a threshold value T is set in the coring circuit 22 for taking out the image high-frequency component Sh of the original image Si as the edge emphasis component Se, and the threshold value T is set as the threshold value. The value control circuit 29 variably controls the original image Si according to the contour component E (= E1 + E2). However, another configuration may be used as described in the contour emphasis processing unit 151 of the second embodiment. .

(Second Embodiment)
FIG. 10 is a block diagram showing the configuration of the contour enhancement processing unit 151 of the second embodiment provided in the image processing unit 14 of the digital camera.

  In the contour emphasis processing unit 151 of the second embodiment, a high frequency component extracting circuit 21 that extracts an image high frequency component Sh from the original image Si, and adding the contour emphasizing component Se to the original image Si, the contour emphasizing image So is obtained. An adder 23 to be obtained, a contour detection circuit (middle region) 24 for detecting the middle band component E1 of the contour components of the original image Si, and a contour detection circuit for detecting the high band component E2 of the contour components of the original image Si ( 25), a coefficient unit 26 that multiplies the mid-band contour component E1 by a predetermined coefficient K1, a coefficient unit 27 that multiplies the high-band contour component E2 by a predetermined coefficient K2, a mid-band contour component E1 and a high-band contour component E2. The adder 28 that obtains the contour component E by addition has the same configuration as that in the contour emphasis processing unit 15 of the first embodiment.

  In the contour emphasis processing unit 151 of the second embodiment, the image high frequency component Sh of the original image Si obtained by the high frequency component extraction circuit 21 is extracted as the contour emphasizing component Se by multiplying the gain G in the coefficient unit 41. The gain G set in the coefficient unit 41 is variably controlled by the gain control circuit 42 according to the contour component E (= E1 × K1 + E2 × K2) of the original image Si.

  The gain control circuit 42 determines that the image component of the target pixel is more likely to be noise as the contour component E of the input target pixel is smaller, and decreases the gain G for the coefficient unit 41. In contrast, as the contour component E of the target pixel to be input is larger, the image component of the target pixel is determined to be a correct contour component, and control gain is set so that the gain G for the coefficient multiplier 41 is increased. .

  Next, a specific operation in the contour emphasis processing unit 151 of the second embodiment provided in the image processing unit 14 of the digital camera having the above configuration will be described.

  That is, the image data Pa having isolated point noise at the center pixel shown in FIG. 8A, and the image data Pb when the vertical contour component Ev1 in the middle region is high at the center pixel shown in FIG. 8B. 8C, when the image data Pc when the high-frequency vertical contour component Ev2 is high in the center pixel is input as the original image Si to the contour emphasis processing unit 151, each original image Si ( As shown in FIG. 9, the image high-frequency component Sh, the mid-frequency contour component E1, the high-frequency contour component E2, and the contour component E for Pa, Pb, and Pc) are all contour enhancement processing of the first embodiment. It is obtained in the same manner as that in the section 15.

  The image high-frequency component Sh for the target pixel of each of the image data Pa, Pb, and Pc is large in the order of Pa> Pc> Pb, whereas the contour component E = E1 + E2 is Pb> Pc> Pa. In this order, the target pixel of the image data Pa can be properly determined to be isolated point noise with a small contour component E even if the image high-frequency component Sh is large. Further, the target pixels of the image data Pb and Pc can be properly determined that the contour component E is large and the image contour even if the image high-frequency component Sh is small.

  In the gain control circuit 42, the contour emphasis component of the image high-frequency component Sh for the coefficient unit 41 increases as the contour component E obtained by adding the middle-frequency contour component E1 and the high-frequency contour component E2 in the target image of the original image Si increases. The extraction gain G as Se is set and controlled to be high, and conversely, the smaller the contour component E obtained by the addition of the middle region contour component E1 and the high region contour component E2 in the target image of the original image Si, Since the extraction gain G as the contour emphasis component Se of the image high-frequency component Sh is controlled to be low, the gains Ga, Gb, and Gc when the image data Pa, Pb, and Pc are respectively input as the original image Si are The setting is controlled as Gb> Gc> Ga according to the order Pb> Pc> Pa of the size of the contour component E.

  Thereby, in the coefficient unit 41, as shown in FIG. 8A, the image high-frequency component Sh of the image data Pa whose target pixel is isolated point noise has an extremely low gain G even if the high-frequency component Sh is large. The image high-frequency components of the image data Pb and Pc when the contour component E of the target pixel is large as shown in FIGS. 8B and 8C are suppressed. Sh is extracted as an edge emphasis component Se amplified by a large gain G even if the high frequency component Sh is small. Therefore, it is possible to obtain a contour-enhanced image So in which only the contour-enhancement component Se of the correct image contour in which the noise component is appropriately suppressed is greatly added to the original image Si.

  Therefore, according to the digital camera having the contour enhancement function of the second embodiment having the above configuration, the image high-frequency component Sh extracted from the original image Si is extracted as the contour enhancement component Se via the coefficient unit 41 in which the gain G is set. In addition, when the contour-enhanced image So is obtained by adding to the original image Si, a middle-region contour component E1 obtained by adding the horizontal contour component Eh1 and the vertical contour component Ev1 in the middle region of the original image Si, and a high-frequency region A contour component E (= E1 + E2) consisting of a high-frequency contour component E2 obtained by adding the horizontal contour component Eh2 and the vertical contour component Ev2 is obtained, and the gain G of the coefficient unit 41 increases as the middle-high frequency contour component E of the original image Si increases. Is controlled to be high, and even if the image high-frequency component Sh is small, it is extracted as a contour emphasizing component Se, and the gain G of the coefficient multiplier 41 is set lower as the middle-high frequency contour component E is smaller. Even if the image high-frequency component Sh is large, the contour emphasizing component Se is extracted with an extremely suppressed value, so that image noise can be prevented from being emphasized and extracted as an image contour, and only the original contour component is increased. A good contour-enhanced image So with less emphasized noise can be obtained.

  Note that, in the contour enhancement processing unit 15 of the first embodiment and the contour enhancement processing unit 151 of the second embodiment, the adder 23 adds the contour enhancement component Se to the original image Si in any of the embodiments. However, the value obtained by adding the contour emphasis component Se to the original image Si in the adder 23 is used as described in the contour emphasis processing unit 152 of the third embodiment. Depending on the size of the contour component E (= E1 + E2) detected from the original image Si, the high-frequency emphasized image Sf is mainly extracted as the contour-enhanced image So. A configuration may be adopted in which the high-frequency suppression image SI obtained by weighted averaging of the original image Si is mainly extracted as the contour emphasis image So.

(Third embodiment)
FIG. 11 is a block diagram showing a configuration of an edge emphasis processing unit 152 of the third embodiment provided in the image processing unit 14 of the digital camera.

  In the contour emphasis processing unit 152 of the third embodiment, the characteristic constituent parts different from the first and second embodiments are any of the contour emphasis processing units 15 and 151 described in the first and second embodiments. However, here, a description will be given of a case where the present invention is configured to be applied to the contour enhancement processing unit 15 of the first embodiment, and is applied to the contour enhancement processing unit 151 of the second embodiment. A description of the configuration will be omitted.

  In the contour emphasis processing unit 152 of the third embodiment, a high frequency component extraction circuit 21 that extracts an image high frequency component Sh from the original image Si, and a threshold T is set for the image high frequency component Sh to obtain the contour emphasizing component Se. A coring circuit 22, an adder 23 for adding a contour emphasis component Se to the original image Si, a contour detection circuit (middle region) 24 for detecting a middle band component E1 among the contour components of the original image Si, A contour detection circuit (high frequency) 25 for detecting a high frequency component E2 of the contour components, a coefficient unit 26 for multiplying the mid frequency contour component E1 by a predetermined coefficient K1, and a coefficient for multiplying the high frequency contour component E2 by a predetermined coefficient K2. 27, a threshold value for variably setting the threshold T for the coring circuit 22 in accordance with a change in the contour component E, an adder 28 for obtaining the contour component E by adding the mid-frequency contour component E1 and the high-frequency contour component E2. In the value control circuit 29 Information, all of which are similar to structure in the edge enhancement processing portion 15 of the first embodiment.

  In the contour emphasis processing unit 152 of the third embodiment, a value obtained by adding the contour emphasis component Se to the original image Si in the adder 23 is extracted as a high-frequency emphasized image (contour emphasized image) Sf, and the original image Si is weighted. The high-frequency-suppressed image (image low-frequency component) SI is acquired through the averaging circuit 50, and the high-frequency emphasized image Sf and the high-frequency-suppressed image SI are mixed by the mixing circuit 51 whose mixing ratio is set and controlled. The image is output as an outline enhanced image So. Then, the mixing ratio between the high frequency emphasized image Sf and the high frequency suppressed image SI set in the mixing circuit 51 is variably controlled by the mixing ratio control circuit 52 according to the contour component E (= E1 + E2) of the original image Si. The configuration.

  The mixture ratio control circuit 52 determines that the smaller the contour component E of the input target pixel is, the higher the possibility that the image component of the target pixel is noise, and the high-frequency-suppressed image (image) for the mixing circuit 51 is determined. The control ratio is set so that the mixing ratio of the low-frequency component) SI is large and the mixing ratio of the high-frequency emphasized image (contour emphasized image) Sf is small, and conversely, the larger the contour component E of the input target pixel is, It is determined that the image component of the target pixel is a correct contour component, the mixing ratio of the high-frequency suppressed image (image low-frequency component) SI to the mixing circuit 51 is small, and the mixing ratio of the high-frequency emphasized image (contour emphasized image) Sf is set. Set the control to be larger.

  The weighted average circuit 50 multiplies the target pixel of the original image Si and the image data of 3 × 3 pixels adjacent to the target pixel by multiplying the weighted average coefficient 21b shown in FIG. An image (image low-frequency component) SI is obtained.

  Next, a specific operation in the contour emphasis processing unit 152 of the third embodiment provided in the image processing unit 14 of the digital camera having the above configuration will be described.

  That is, the image data Pa having isolated point noise at the center pixel shown in FIG. 8A, and the image data Pb when the vertical contour component Ev1 in the middle region is high at the center pixel shown in FIG. 8B. 8C, when the image data Pc when the high-frequency vertical contour component Ev2 is high at the center pixel is input as the original image Si to the contour emphasis processing unit 152, each original image Si ( As shown in FIG. 9, the image high-frequency component Sh, the mid-frequency contour component E1, the high-frequency contour component E2, and the contour component E for Pa, Pb, and Pc) are all contour enhancement processing of the first embodiment. It is obtained in the same manner as that in the section 15.

  The image high-frequency component Sh for the target pixel of each of the image data Pa, Pb, and Pc is large in the order of Pa> Pc> Pb, whereas the contour component E = E1 + E2 is Pb> Pc> Pa. In this order, the target pixel of the image data Pa can be properly determined to be isolated point noise with a small contour component E even if the image high-frequency component Sh is large. Further, the target pixels of the image data Pb and Pc can be properly determined that the contour component E is large and the image contour even if the image high-frequency component Sh is small.

  First, in the threshold value control circuit 29, the higher the contour component E obtained by adding the middle region contour component E1 and the high region contour component E2 in the target image of the original image Si, the larger the image high region component for the coring circuit 22 is. The extraction threshold T as the Sh contour emphasis component Se is controlled to be low, and conversely, the smaller the contour component E obtained by adding the middle region contour component E1 and the high region contour component E2 in the target image of the original image Si, the smaller the contour component E is. Since the extraction threshold value T as the contour emphasis component Se of the image high-frequency component Sh for the coring circuit 22 is set to be high, the threshold values Ta when the image data Pa, Pb, and Pc are respectively input as the original image Si are controlled. , Tb, Tc are set and controlled in the order of Ta> Tc> Tb in inverse proportion to the order Pb> Pc> Pa of the size of the contour component E.

  Thereby, in the coring circuit 22, the image high frequency component Sh of the image data Pa in which the target pixel is isolated point noise as shown in FIG. 8A is removed even if the high frequency component Sh is large. The image high frequency component Sh of the image data Pb and Pc when the contour component E of the target pixel is large as shown in FIGS. 8B and 8C is not extracted as the contour emphasizing component Se. Even if the component Sh is small, it is extracted as the contour emphasizing component Se without being removed. At this stage, only the contour emphasizing component Se of the correct image contour from which the noise component is properly removed is added to the original image Si. A region-enhanced image (outline-enhanced image) Sf is obtained.

  In the mixing ratio control circuit 52, as the contour component E (= E1 + E2) of the original image Si is larger, the mixing ratio of the high frequency emphasized image (contour emphasized image) Sf to the mixing circuit 51 is set and controlled. As the contour component E (= E1 + E2) of the original image Si is smaller, the ratio of the high-frequency emphasized image (contour emphasized image) Sf to the mixing circuit 51 is lower, and the mixing ratio of the high-frequency suppressed image (image low-frequency component) SI is set. Since the setting control is high, the mixing ratios Ma, Mb, and Mc of the high-frequency emphasized image Sf when the image data Pa, Pb, and Pc are respectively input as the original image Si have the size of the contour component E. The setting is controlled in the order of Tb> Tc> Ta according to the order Pb> Pc> Pa.

  Thereby, when the contour component E (= E1 + E2) of the original image Si is large, the contour-enhanced image So having a high mixing ratio of the high-frequency emphasized image Sf is obtained, and conversely, the contour component E (= When E1 + E2) is small, a contour-enhanced image So having a high mixing ratio of the high-frequency-suppressed image SI can be obtained, and not only a contour-enhanced image So having a correct image contour in which noise components are appropriately suppressed can be obtained. , An edge-enhanced image So with improved SN ratio can be obtained.

  Therefore, according to the digital camera having the contour enhancement function of the third embodiment having the above-described configuration, the image high-frequency component Sh extracted from the original image Si is extracted as the contour enhancement component Se by the coring circuit 22 in which the threshold value T is set. The high-frequency emphasized image (contour emphasized image) Sf is obtained by adding to the original image Si, and the high-frequency emphasized image Sf and the high-frequency suppressed image SI obtained by weighted averaging of the original image Si are mixed. When the contour-enhanced image So is obtained, the middle contour component E1 obtained by adding the horizontal contour component Eh1 and the vertical contour component Ev1 in the middle region of the original image Si, and the horizontal contour component Eh2 and the vertical contour component in the high region. A contour component E (= E1 + E2) consisting of a high frequency contour component E2 obtained by adding Ev2 is obtained, and when the middle high frequency contour component E of the original image Si is large, the threshold of the coring circuit 22 is obtained. Even if T is set to a low value and the image high frequency component Sh is small, a high frequency emphasized image (contour emphasized image) Sf extracted as the contour emphasizing component Se is obtained. The threshold value T of 22 is set to a high value to obtain a high-frequency emphasized image (contour-enhanced image) Sf that is not extracted as the contour-enhanced component Se even if the image high-frequency component Sh is large. Further, when the middle and high frequency contour component E of the original image Si is large, a contour emphasized image So having a high mixing ratio of the high frequency emphasized image (contour emphasized image) Sf is obtained, and the middle and high frequency contour component of the original image Si is obtained. When E is small, an edge-enhanced image So in which the mixing ratio of the high-frequency emphasized image (contour-enhanced image) Sf is low and the mixing ratio of the high-frequency-suppressed image SI is high is obtained. Can be prevented and taken out, and a good contour-enhanced image So with little noise in which only the original contour component is greatly enhanced can be obtained. In addition, it is possible to obtain a contour-enhanced image So with an improved SN ratio.

  In the description of the contour emphasis processing units 15, 151, and 152 in each of the above embodiments, the description has been given mainly on the case where each unit of the contour emphasis processing is implemented by a hardware configuration, but for example, FIG. 2, FIG. 4, FIG. By replacing the block configuration diagrams shown in FIG. 10 and FIG. 11 as functional blocks when implemented by a computer program, it is possible to realize contour emphasis processing based on program control by a control unit (CPU) 16 that is a computer.

  In other words, by installing the contour enhancement processing program described in each of the embodiments on a computer having an image processing function such as a personal computer, the contour enhancement processing for various image data as in the above embodiments is performed. And the same effect can be obtained.

  The present invention is not limited to the above-described embodiments, and the embodiments can be variously modified without departing from the scope of the invention. Further, each of the embodiments includes inventions at various stages, and various inventions can be extracted by appropriately combining a plurality of disclosed constituent elements. For example, even if some constituent requirements are deleted from all the constituent requirements shown in each embodiment or some constituent features are combined, the problems described in the column of the problem to be solved by the invention can be solved. When the effects described in the column of the effect of the invention can be obtained, a configuration in which these constituent elements are deleted or combined can be extracted as an invention.

1 is a block diagram showing a configuration of a digital camera having a contour enhancement function according to an embodiment of an image processing apparatus of the present invention. The block diagram which shows the structure of the outline emphasis process part 15 of 1st Embodiment with which the image process part 14 of the said digital camera was equipped. It is a figure which shows an example of the filter coefficient used in the high frequency component extraction circuit 21 of the said outline emphasis processing part 15, The same figure (A) shows the image high frequency component Sh directly by multiplying with the original image Si. FIG. 5B is a diagram showing a Laplacian coefficient 21a for obtaining the image low-frequency component by subtracting the image low-frequency component SI from the original image Si to obtain the image high-frequency component Sh and multiplying the original image Si. The figure which shows the weighted average coefficient 21b for obtaining SI. The block diagram which shows the structure of other embodiment of the high frequency component extraction circuit 21 in the said outline emphasis processing part 15. FIG. The block diagram which shows the structure of the outline detection circuit 24 (25) in the said outline emphasis process part 15. FIG. It is a figure which shows the operator coefficient for outline component detection used in the 1st outline detection circuit (middle area) 24 of the said outline emphasis processing part 15, The figure (A) is a horizontal outline component (middle area) detection coefficient. FIG. 24B is a diagram showing a vertical contour component (middle region) detection coefficient 24v. It is a figure which shows the operator coefficient for the contour component detection used in the 2nd contour detection circuit (high region) 25 of the said contour emphasis processing part 15, The figure (A) is a horizontal contour component (high region) detection coefficient. FIG. 25B is a diagram illustrating a vertical contour component (high frequency) detection coefficient 25v. It is a figure which shows the specific example of the image data input into the said outline emphasis process part 15 as original image Si, and the same figure (A) is a figure which shows the image data Pa in case there exists an isolated point noise in a center pixel. (B) is a diagram showing image data Pb when the central vertical contour component Ev1 is high at the center pixel, and (C) shows image data Pc when the high frequency vertical contour component Ev2 is high at the central pixel. Figure. When the image data Pa, Pb, and Pc shown in FIG. 8 are input as the original image Si to the contour enhancement processing unit 15, the image high-frequency component Sh, the mid-region contour component E1, and the high The figure which shows the area | region outline component E2 and the outline component E in contrast. The block diagram which shows the structure of the outline emphasis process part 151 of 2nd Embodiment with which the image process part 14 of the said digital camera was equipped. The block diagram which shows the structure of the outline emphasis processing part 152 of 3rd Embodiment with which the image processing part 14 of the said digital camera was equipped.

Explanation of symbols

10 ... Imaging unit (CCD)
11 ... A / D converter 12 ... Image interpolation circuit (RGB)
DESCRIPTION OF SYMBOLS 13 ... Matrix circuit 14 ... Image processing part 15 ... Outline emphasis part (1st Embodiment)
151... Outline emphasis unit (second embodiment)
152. Outline emphasis unit (third embodiment)
16: Control unit (CPU)
DESCRIPTION OF SYMBOLS 17 ... Image memory 18 ... Display buffer 19 ... Display part 21 ... High region component extraction circuit 22 ... Coring circuit 23, 28, 34 ... Adder 24 ... 1st outline detection circuit (middle region)
25 ... Second contour detection circuit (high frequency range)
26, 27 ... Coefficient unit 29 ... Threshold control circuit 30 ... Contour detection circuit (horizontal)
31 ... Contour detection circuit (vertical)
32, 33... Absolute value circuit 41... Coefficient unit (G)
42 ... Gain (G) control circuit 50 ... Weighted average circuit 51 ... Mixing circuit 52 ... Mixing ratio control circuit

Claims (6)

High-frequency component extraction means for extracting high-frequency components of the image;
Contour enhancement component generating means for generating a contour enhancement component from the high frequency component;
Addition means for acquiring a contour-enhanced image in which the contour on the image is enhanced by adding the contour-enhanced component to the image; and
The outline emphasis component generation means includes
Coring means for outputting, as the contour emphasis component, a component consisting only of a target pixel equal to or higher than a predetermined threshold among the target pixels constituting the high-frequency component;
Middle region contour component detecting means for detecting a contour component of the middle region of the image;
High frequency contour component detecting means for detecting a high frequency contour component of the image;
For the target pixel, calculation means for calculating a contour component value based on the contour component of the middle region and the contour component of the high region,
When the contour component value is larger than a predetermined value, the threshold value is set to a first value, and when the contour component value is smaller than the predetermined value, the threshold value is set to a second value higher than the first value. Threshold setting means for setting to
An image processing apparatus. High-frequency component extraction means for extracting high-frequency components of the image;
Based on the contour component of the middle region and the contour component of the high region, contour enhancement component generation means for generating a contour enhancement component from the high region component;
Addition means for acquiring a contour-enhanced image in which the contour on the image is enhanced by adding the contour-enhanced component to the image; and
The outline emphasis component generation means includes
Amplifying means for outputting a component obtained by amplifying each target pixel constituting the high-frequency component with a predetermined gain as the contour enhancement component;
Middle region contour component detecting means for detecting a contour component of the middle region of the image;
High frequency contour component detecting means for detecting a high frequency contour component of the image;
For the target pixel, calculation means for calculating a contour component value based on the contour component of the middle region and the contour component of the high region
Gain control means for increasing the gain as the contour component value increases and decreasing the gain as the contour component value becomes smaller than the predetermined value.
An image processing apparatus. High-frequency component extraction means for extracting high-frequency components of the image;
Contour enhancement component generating means for generating a contour enhancement component from the high frequency component;
Adding means for obtaining a high-frequency emphasized image by adding the contour enhancement component to the image;
Low-frequency component extracting means for extracting a low-frequency component of the image;
A mixing unit that outputs an outline-enhanced image by mixing the high-frequency emphasized image and the low-frequency component at a predetermined mixing ratio;
Middle region contour component detecting means for detecting a contour component of the middle region of the image;
High frequency contour component detecting means for detecting a high frequency contour component of the image;
For the target pixel, calculation means for calculating a contour component value based on the contour component of the middle region and the contour component of the high region
The mixing ratio is set so that the proportion of the high-frequency emphasized image in the contour-enhanced image increases as the contour component value increases, and the high-frequency emphasized image in the contour-enhanced image decreases as the contour component value decreases. Mixing ratio control means for setting the mixing ratio so that the ratio of
An image processing apparatus comprising: The mid-range contour component detecting means is
Mid-horizontal contour component detecting means for detecting a horizontal contour component of the mid-range of the image;
Mid-range vertical contour component detection means for detecting a vertical contour component of the mid-range of the image;
A mid-region contour component adding means for obtaining the contour component by adding the horizontal contour component of the mid-region and the vertical contour component of the mid-region, and
The high frequency contour component detecting means is
High-frequency horizontal contour component detecting means for detecting a high-frequency horizontal contour component of the image;
High frequency vertical contour component detection means for detecting a high frequency vertical contour component of the image;
High-frequency contour component adding means for acquiring the contour component by adding the high-frequency horizontal contour component and the high-frequency vertical contour component;
The image processing apparatus according to claim 1, wherein the image processing apparatus is an image processing apparatus. The mid-range horizontal contour component detecting means is
A total sum of values obtained by multiplying each of the target pixel located at the center of the region of interest and the pixel adjacent to the target pixel by the mid-range horizontal coefficient corresponding to the pixel position is detected as the horizontal contour component of the mid-range. ,
The mid-range horizontal coefficient is
The total value of each coefficient is 0, and the coefficient corresponding to the position of the target pixel and the coefficient corresponding to the position of each pixel adjacent to the target pixel in the horizontal direction are each 0.
The middle vertical contour component detecting means is
The sum of values obtained by multiplying each of the target pixel located at the center of the region of interest and the pixel adjacent to the target pixel by the middle vertical coefficient corresponding to the pixel position is detected as the vertical contour component of the middle region. ,
The mid-range vertical coefficient is
The total value of each coefficient is 0, and the coefficient corresponding to the position of the target pixel and the coefficient corresponding to the position of each pixel adjacent to the target pixel in the vertical direction are each 0.
The image processing apparatus according to claim 4. The high-frequency horizontal contour component detection means is
The sum of values obtained by multiplying the target pixel located at the center of the target region and the pixel adjacent to the target pixel by the high frequency horizontal coefficient corresponding to the pixel position is detected as the horizontal contour component of the high frequency. ,
The high frequency horizontal coefficient is
The total value of each coefficient is 0, and the coefficient corresponding to the position of the target pixel and the coefficient corresponding to the position of each pixel adjacent to the target pixel in the horizontal direction are positive values.
The high frequency vertical contour component detecting means is
The sum of values obtained by multiplying the target pixel located at the center of the target region and the pixel adjacent to the target pixel by the high frequency vertical coefficient corresponding to the pixel position is detected as the vertical contour component of the high frequency. ,
The high frequency vertical coefficient is
The total value of each coefficient is 0, and the coefficient corresponding to the position of the target pixel and the coefficient corresponding to the position of each pixel adjacent to the target pixel in the vertical direction are positive values.
The image processing apparatus according to claim 4.

Images