JP2006004124A - Picture correction apparatus and method, and picture correction program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem in that a noise at a specific picture portion encountered after brightness correction application is noticeable. <P>SOLUTION: A face region partition circuit 11 partitions a face picture included in a picture. An average brightness value calculation circuit 13 calculates the average brightness value of the face picture. In addition, a noise amount calculation circuit 12 calculates a noise amount included in face picture data. When the noise amount has an acceptable noise value or below, a target average brightness value inputted by a user is assigned to the correction value of the average brightness value. A correction coefficient calculation circuit 15 calculates a correction coefficient α based on the relation. When the noise amount exceeds the acceptable noise amount, a new target brightness value is decided so that the noise amount included in the picture data representing the brightness-corrected face picture may agree with the acceptable noise amount. The newly decided target brightness value is assigned to the correction value of the average brightness value. The correction coefficient α is calculated based on the relation. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image correction apparatus and method for performing image correction on digital image data, and a program for causing a computer to perform image correction.

When creating a color print of digital image data obtained by reading an image recorded on a photographic negative film or reversal film with a scanner, or an image represented by digital image data obtained by a digital camera, etc. Brightness correction (brightness correction, brightness correction) for digital image data so that the brightness (brightness, density) of the desired subject image part such as a human face included in the image is an appropriate brightness (brightness, density). Density correction) may be performed. In Patent Document 1, a person's face included in an image is extracted, and density correction is performed on the image data so that the face area average density is within the correction target range.
JP 2000-196890 A

  However, when the image is corrected brightly, the noise included in the image tends to be visually recognized more clearly than before correction. That is, as in Patent Document 1, the density is uniformly distributed over the entire image so that the average density (face area average density) of a desired subject image portion such as a human face included in the image becomes the correction target range. When correction (correction that makes it brighter) is performed, if the desired subject image portion in the image contains a lot of noise, the desired subject image portion will become brighter due to the density correction, but the noise will become conspicuous. End up. If a portion other than the desired subject image portion contains noise, the noise will also be noticeable in that portion.

  When image data to be processed contains noise, it is possible to remove noise components in the image data by using a noise removal filter or the like. However, complete noise removal is difficult. Further, when noise removal processing is performed on image data having a large resolution (a large number of pixels when displayed), a long processing time is required, and the resources of the computer apparatus are occupied for a long time.

  Therefore, the present invention has been made in view of the above circumstances. Even if noise is included in an image portion to be noticed in an image, the present invention pays attention after performing brightness (density, luminance) correction. The purpose is to prevent the noise in the power image portion from becoming noticeable.

  The present invention provides an apparatus and a method for correcting the brightness (luminance) of at least a part of given image data. In the present invention, a luminance value is used as data representing the brightness of an image. Since the luminance value can be uniquely converted into a density value, the luminance value can be read as the density value in the present invention.

  An image correction apparatus according to the present invention includes a representative luminance value calculating means for calculating a representative luminance value in a specific image included in an image represented by given image data, and an amount of noise included in the specific image data representing the specific image. The calculated noise amount calculating means, the target luminance value determining means for determining the target luminance value according to the noise amount calculated by the noise amount calculating means, and the representative luminance value calculated by the representative luminance value calculating means are the target luminance value. Based on the data representing the luminance conversion line or the luminance conversion curve and the data representing the created luminance conversion line or the luminance conversion curve so as to be corrected to the target luminance value determined by the luminance value determining means. , And a brightness correction means for correcting the brightness of at least the specific image data representing the specific image.

  The image correction method according to the present invention calculates a representative luminance value in a specific image included in an image represented by given image data, calculates a noise amount included in the specific image data representing the specific image, and calculates A target brightness value is determined according to the amount of noise, and data representing a brightness conversion line or a brightness conversion curve is created so that the calculated representative brightness value is corrected to the determined target brightness value. Alternatively, brightness correction is performed on at least specific image data representing the specific image based on data representing a luminance conversion curve.

  The digital image data given is digital image data obtained by reading an image recorded on a photographic negative film or a reversal film with a scanner, a digital image obtained by a digital camera, a mobile phone with a camera, a PDA with a camera, or the like. Includes data. Digital image data obtained by a scanner, digital camera, camera-equipped mobile phone, camera-equipped PDA, etc. is given to the image correction device.

  The specific image means an image portion (image region) having a specific structure, shape, color, etc. included in an image represented by given image data. The specific image is not limited to the partial image area included in the image represented by the given image data, but includes that the entire image represented by the given image data is the specific image. Specific images include, for example, human faces, torso, hands, feet, etc., faces of animals other than humans (cats, dogs, monkeys, chickens, rabbits, snakes, etc.), plants (cherries, roses, plums, etc.) Examples include images representing petals, leaves, main objects in the background (Mt. Fuji, moon, temples, etc.), structures (automobiles, etc.), and the like. Since these specific images have a specific structure, shape, color, etc., image data (specific image data) representing the specific image is detected (partitioned) from given image data (image data representing the entire image). , Extract).

  The representative luminance value in the specific image is a value representative of the luminance value in the specific image. In one embodiment, the average value of the luminance values of a plurality of pixels constituting the specific image is used as the representative luminance value. Can do. Of course, the median value and the mode value may be used as the representative luminance value.

  Digital image data includes noise caused by electronic components mounted inside the camera, noise generated when reading images recorded on photographic negative and reversal films with a scanner, noise captured by a lens, Noise or the like generated by image processing may be added. The amount of noise included in the specific image data is calculated by the noise amount calculation means.

  A target luminance value is determined in accordance with the amount of noise included in the calculated specific image data. That is, the determined target brightness value varies according to the amount of noise included in the specific image data.

  Data representing a luminance conversion line or luminance conversion curve is created so that the calculated representative luminance value is corrected to the determined target luminance value, and based on the created data representing the luminance conversion line or luminance conversion curve. Thus, at least the specific image data representing the specific image is subjected to luminance correction.

  The luminance conversion curve or data representing the luminance conversion curve is created so that the representative luminance value is corrected to the target luminance value. That is, the pixel having the representative luminance value obtained from the specific image data included in the image data before the luminance correction has the target luminance value by the luminance correction processing based on the data representing the luminance conversion line or the luminance conversion curve to be created. It is corrected as follows.

  For pixels having a luminance value other than the representative luminance value, the luminance is corrected based on the created luminance conversion line or data representing the luminance conversion curve. The luminance conversion line or luminance conversion curve is created based on the correspondence between the representative luminance value and the target luminance value. For example, in the graph representing the input / output relationship between the input luminance value and the output luminance value, the intersection of the representative luminance value (which is the input luminance value) and the target luminance value (which is the output luminance value) and the origin (input luminance value “ A luminance conversion straight line passing through an output luminance value corresponding to “0” (“0”) can be created. A luminance conversion curve passing through three points of the intersection of the representative luminance value and the target luminance value, the origin, and the corresponding point between the maximum luminance values (for example, the output luminance value corresponding to the input luminance value “255” is “255”). It can also be created. In any case, the created data representing the luminance conversion line or luminance conversion curve defines not only the correspondence between the representative luminance value and the target luminance value, but also the correspondence between other input luminance values and output luminance values. To do.

  According to the present invention, the target luminance value is determined in accordance with the amount of noise in the specific image included in the image represented by the given image data (included in the specific image data). The correction value is a luminance value. Since the representative luminance value (correction value) is calculated according to the noise amount, correction of the luminance value (density and brightness) according to the noise amount is realized. Since the luminance value is corrected in consideration of the amount of noise, the luminance value (density and brightness) of the specific image is corrected so that noise in the corrected image (particularly the specific image) is not noticeable. be able to.

  In one embodiment, the image correction apparatus includes desired target luminance value input means for receiving an input of a desired target luminance value, and the representative luminance value calculated by the representative luminance value calculating means is input by the target luminance value input means. A predicted noise amount calculating means for calculating a predicted noise amount that will be included in the specific image data after the brightness correction based on the desired target brightness value and the noise amount calculated by the noise amount calculating means; The target luminance value determining means determines the target luminance value according to the predicted noise amount calculated by the predicted noise amount calculating means.

  Instead of the noise amount in the specific image included in the image represented by the given image data, the noise amount (predicted noise amount) in the specific image when the luminance correction is assumed is used to determine the target luminance value. By determining the target luminance value using the predicted noise amount and creating data representing a luminance conversion line or luminance conversion curve based on the target luminance value using the determined predicted noise amount, the noise in the specific image is It can be made inconspicuous after brightness correction.

  Preferably, the target luminance value determining means is set or inputted in advance, an allowable noise amount for noise that will be included in the specific image data after luminance correction, a calculated noise amount, or a calculated noise amount Based on the predicted noise amount, a target luminance value is determined such that the noise amount or the predicted noise amount is less than or equal to the allowable noise amount. An allowable noise amount can be set or input, and the target luminance value is determined according to the allowable noise amount, so that luminance correction is performed according to the tolerance of the user's noise.

  In one embodiment, the target luminance value determining means uses the desired target luminance value input from the desired target luminance value input means as the target luminance value when the calculated predicted noise amount is less than or equal to the allowable noise amount. If the calculated predicted noise amount exceeds the allowable noise amount, a new target luminance value that makes the predicted noise amount equal to the allowable noise amount is set to the representative luminance value, allowable noise amount, and Calculation is performed based on the amount of noise, and the calculated new target luminance value is determined as the target luminance value. When the calculated predicted noise amount is less than or equal to the allowable noise amount, luminance correction as desired by the user is realized. Further, since the specific image after the luminance correction always has a noise amount equal to or less than the allowable noise amount, the user's request for noise is surely satisfied. If the allowable noise amount is exceeded, a new target luminance value that makes the predicted noise amount equal to the allowable noise amount is calculated based on the representative luminance value, the allowable noise amount, and the noise amount. Thus, it is possible to calculate a target luminance value that does not change the representative luminance value as much as possible while reliably satisfying the user's request for noise.

  The present invention also provides a program for causing a computer to execute image correction processing (a program for causing a computer to function as an image correction apparatus).

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of the digital print system in the first embodiment. The digital print system includes an image correction device 1 and peripheral devices (input device 2, display device 3, storage device 4, and printer 5) connected to the image correction device 1. FIG. 2 is a block diagram showing a detailed electrical configuration of the image correction apparatus 1 which is a core apparatus of the digital printing system, together with a data flow. In FIG. 2, the display device 3 and the printer 5 are omitted from the peripheral devices shown in FIG.

  In the image correction apparatus 1, the brightness (luminance) of a portion (hereinafter referred to as a face image portion) representing a person's face included in an image represented by given image data conforms to a set target luminance value. Luminance correction is performed on image data so as to have brightness (luminance). As will be described later, the image correction apparatus 1 determines the luminance when the predicted noise amount that will be included in the image data representing the face image portion after the luminance correction exceeds (is expected to exceed) the predetermined noise amount. A new target luminance value is determined so that the predicted noise amount that will be included in the image data representing the corrected face image portion does not exceed a predetermined noise amount.

  An input device 2 (keyboard, mouse, etc.) connected to the image correction apparatus 1 is used for inputting a target average luminance value, an allowable noise amount, etc. (the target average luminance value and the allowable noise amount will be described later). On the display screen of the display device 3, an image or the like is displayed based on a screen for setting a target average luminance value input from the input device 2 and image data before and after correction. Image data is stored in the storage device (hard disk, memory card, CD-ROM, etc.) 4, and image correction processing is performed in the image correction device 1 on the image data read from the storage device 4. The printer 5 prints an image represented by the corrected image data on photographic paper or the like.

  The image processing apparatus 1 includes a face area partition circuit 11, a noise amount calculation circuit 12, an average luminance value calculation circuit 13, a target average luminance value adjustment circuit 14, a correction coefficient calculation circuit 15, and an image correction circuit 16.

  The image correction circuit 16 performs luminance correction on the image data read from the storage device 4 connected to the image correction device 1. The luminance correction processing performed in the image correction circuit 16 follows a correction function that defines the correspondence between the input luminance value and the output luminance value (correction value). A coefficient of a correction function used for luminance correction in the image correction circuit 16 is calculated in the correction coefficient calculation circuit 15.

  First, the processing of the face area dividing circuit 11, the noise amount calculating circuit 12, and the average luminance value calculating circuit 13 will be described. The face area dividing circuit 11, the noise amount calculating circuit 12, and the average luminance value calculating circuit 13 perform processing for calculating the amount of noise included in the face image portion and processing for calculating the average luminance value of the face image portion.

  It is assumed that the image represented by the image data to be corrected includes an image representing a human face (face image). Image data representing an image including a face image is read from the storage device 4.

  Image data read from the storage device 4 (hereinafter referred to as input image data) is input to the face area dividing circuit 11, the noise amount calculating circuit 12, and the average luminance calculating circuit 13, respectively.

  The face area partitioning circuit 11 partitions a face image portion included in an image represented by input image data (hereinafter referred to as an input image) from other image portions (detects a face image portion and defines a boundary. ) Circuit.

  For the process of partitioning the face image portion included in the input image, various conventional or novel section (detection) techniques can be used. For example, a method of decomposing an input image into a plurality of regions and partitioning (detecting) a face image portion using a two-dimensional histogram of hue and saturation obtained for each decomposed image (Japanese Patent Laid-Open No. 5-100328) ), A method of segmenting (detecting) the face image portion from other image portions by pattern matching (Japanese Patent Laid-Open No. 8-122944), detecting a skin candidate region in the input image, and detecting the main in the detected skin candidate region A method of segmenting (detecting) a face image portion by detecting various facial features (eyes, eyebrows, hair, nose and mouth) (Japanese Patent Laid-Open No. 2002-203239), setting a learning frame in the input image, Learning image data representing an image portion defined by the learning frame, extracting one or more representative feature vectors representing the learning frame, setting a search frame in the image, A method for extracting a face image portion from a search frame based on the similarity between each of the representative feature vectors and a plurality of feature vectors extracted from image data representing an image portion defined by the search frame (Japanese Patent Application 2002-2000) 37889).

  The face area dividing circuit 11 includes 1 (or 0) for each pixel included in the sectioned (detected) face image part (in the face image area) and an image part other than the face image part (outside the face image area). Binary data (data in which 1 or 0 is associated with each pixel address) with each pixel set to 0 (or 1) is output. 3 (A) shows the input image represented by the input image data, FIG. 3 (B) shows the face image portion divided (detected) by the face area dividing circuit 11, and FIG. 3 (C) shows the face image. Binary data (mask data) output from the face area dividing circuit 11 with each pixel included in the portion being 1 and each pixel included in the image portion other than the face image portion being 0 is schematically illustrated. Show.

  The binary data output from the face area dividing circuit 11 indicates the position and range (area position) of the face image portion included in the input image. Hereinafter, the binary data output from the face area dividing circuit 11 is referred to as face area information.

  The face area information output from the face area dividing circuit 11 is input to the noise amount calculation circuit 12 and the average luminance value calculation circuit 13, respectively.

  The noise amount calculation circuit 12 calculates the noise amount of the face image portion included in the input image. As described above, the input image data representing the input image including the face image is given from the recording device 4 to the noise amount calculation circuit 12. The area position of the face image portion included in the input image is specified by the face area information output from the face area dividing circuit 12.

  In general, image noise appears as a component in a high frequency region. The noise amount calculation circuit 12 performs filtering with a high-pass filter on each pixel constituting the face image portion, and obtains a high frequency region component of each pixel constituting the face image portion. For example, a 3 × 3 pixel filter shown in FIG. 4 is used for the high-pass filtering.

The variance value σ ² is calculated using the high frequency region components obtained for each of the pixels constituting the face image portion. The calculated variance value σ ² is treated as the noise amount of the face image portion included in the input image.

  The average luminance value calculation circuit 13 calculates an average value of luminance values for each pixel constituting the face image portion included in the input image. The calculated average value is an average luminance value (Yave). The average luminance value (Yave) is calculated by the following formula 1.

  (ΣYi) / n (1)

  Here, Yi represents the luminance value for each pixel constituting the face image portion, and n represents the number of pixels in the face image portion.

  Instead of the average value, the median value or the mode value may be used as the average luminance value (Yave).

  When the image data given to the average luminance value calculation circuit 13 is expressed by RGB values (true values) for each pixel, prior to calculating the average luminance value (Yave), A luminance value Y for each pixel is calculated.

  Y = 0.2126 ・ R + 0.7152 ・ G + 0.0722 ・ B Formula 2

Next, the processing of the target average luminance value determination circuit 14 will be described. The target average luminance determination circuit 14 determines a target average luminance value to be given to the correction coefficient calculation circuit 15 (hereinafter referred to as a determined target average luminance value ( Ytarget )). The target average luminance value ( Ytarget ) determined by the target average luminance value determining circuit 14 is different depending on the noise amount σ ² of the face image portion calculated by the noise amount calculating circuit 12, as will be described below. become.

As described above, the operator of the digital printing system uses the input device 2 to input the target average luminance value (Ytarget) and the allowable noise amount σ ² (maximum noise amount that is not handled when noise is included). The The inputted target average luminance value (Ytarget) and allowable noise amount σ ² are given to the target average luminance value determining circuit 14. Further, the average luminance value (Yave) calculated by the average luminance value calculation circuit 13 is also input to the target average luminance value determination circuit 14.

The target average luminance value determination circuit 14 calculates a determined target average luminance value ( Ytarget ) (correction value of the average luminance value (Yave)) by either one of the following formulas 3 and 4.

Determination target average luminance value ( Ytarget ) = target average luminance value (Ytarget)
(When σ ≤ σ (Yave / Ytarget)) ・ ・ Equation 3

Determination target average luminance value ( Ytarget ) = average luminance value (Yave) · σ / σ
(when σ> σ · (Yave / Ytarget)) · · · Equation 4

Here, sigma is the amount of noise sigma ² of the square root of the face image, sigma represents the allowable amount of noise sigma ² of the square root respectively. The amount of noise sigma ² square root sigma and the square root of the allowable noise quantity sigma ² sigma of the face image, both of which are used as a value representing the magnitude of the noise amount.

  In Expression 3 and Expression 4, the expression including the inequality sign in () is used to determine whether the amount of noise that will be included in the face image portion when the luminance is corrected exceeds the allowable noise amount. Used. An expression including an inequality sign will be described.

When the image is brightly corrected (the luminance value of the image data is corrected to a high value), the noise recognized in the corrected image becomes large. When the image is corrected to be dark, noise recognized in the corrected image is reduced. Given the variation in the size and noise of the luminance correction is proportional to the predicted value of the noise amount of the face image, assuming that the luminance correction: If the (predicted noise amount prediction variance) was sigma ^2, predicted noise The square root Σ of the quantity Σ ² is expressed by Equation 5 below.

Square root of predicted noise Σ
= (Target average luminance value (Ytarget) / Average luminance value (Yave)) · Square root of the amount of noise in the face image part σ
.... Formula 5

Considering that the square root Σ of the predicted noise amount Σ ² of the face image portion when the luminance is corrected is set to be equal to or less than the square root σ of the allowable noise amount σ ² described above, the following inequality is established.

Square root of predicted noise amount Σ ≦ square root of allowable noise amount σ ² σ Equation 6

  When Expression 5 and Expression 6 are combined into one expression, the conditional expression in parentheses in Expression 3 is established. In accordance with the conditional expression in () in Expression 3, the conditional expression in () in Expression 4 is satisfied.

Simply, by comparing the noise amount σ ² (or its square root σ) of the face image portion calculated by the noise amount calculation circuit 12 with the set allowable noise amount σ ² (or its square root σ), it is also possible to determine whether the face image portion of the amount of noise sigma ² (or its square root sigma) exceeds the amount of noise sigma ² should be allowed (or its square root sigma) (of course, be determined so Good). Alternatively, the predicted noise amount Σ ² (or the square root Σ) of the face image portion described above and the set allowable noise amount σ ² (or the square root σ) thereof are compared to thereby calculate the predicted noise amount Σ of the face image portion. ² (or its square root Σ) can also be determined (or, of course, can be determined) whether it exceeds the allowable noise amount σ ² (or its square root σ). By using the conditional expression established based on the above-described Expression 5 and Expression 6, the square root Σ of the predicted noise amount Σ ² of the face image portion when the luminance is corrected is equal to or less than the square root σ of the allowable noise amount σ ^2. condition (formula 6), the noise amount sigma ² of the square root of the face image sigma, the square root of the allowable noise quantity sigma ² sigma, be determined using the average luminance value (Yave) and the target average luminance value (Ytarget) it can. A relatively accurate determination of the amount of noise is made by paying attention to the amount of noise in the face image portion after luminance correction.

When it is determined that the noise amount of the face image portion predicted after the luminance correction is equal to or less than the allowable noise amount (the inequality of Equation 3 is established), the target average luminance value input from the input device 2 (Ytarget) is the determined target average luminance value ( Ytarget ) (Formula 3). The determined target average luminance value ( Ytarget ) (the same luminance value as the input target average luminance value (Ytarget)) is output from the target average luminance value determining circuit 14.

When it is determined that the amount of noise of the face image portion predicted after luminance correction exceeds the amount of noise that should be allowed (the inequality of Equation 4 is established), the input is made from the input device 2 based on Equation 4. A determined target average luminance value ( Ytarget ) having a value different from the target average luminance value ( Ytarget ) is calculated.

Equation 4 is an equation obtained by substituting the right side of Equation 6 for the left side of Equation 5. According to Equation 4, the noise amount in the corrected face image portion is the same as the noise amount to be allowed (limit noise). The determined target average luminance value ( Ytarget ) is calculated. The determined target average luminance value ( Ytarget ) calculated when the conditional expression in parentheses in (4) is satisfied is smaller than the input target average luminance value (Ytarget). To what extent the determined target average luminance value ( Ytarget ) is smaller than the input target average luminance value (Ytarget) is determined by the preset allowable noise amount σ ² and the noise amount calculation circuit 12. It varies depending on the amount of noise σ ² included in the calculated face image portion. In general, for example, when the correction is performed so that the face image portion is brightened, an allowable noise amount σ ² that does not become as small as possible from the input target average luminance value (Ytarget) will be set.

The calculated luminance value is output from the target average luminance value determining circuit 14 as the determined target average luminance value ( Ytarget ).

The determined target average luminance value ( Ytarget ) output from the target average luminance value determining circuit 14 is given to the correction coefficient calculating circuit 15. The correction coefficient calculation circuit 15 calculates a coefficient (correction coefficient α) of a relational expression between the input luminance value and the output luminance value (corrected luminance value) based on the following Expression 7.

Correction coefficient α = determined target average luminance value ( Ytarget ) / average luminance value (Yave) Equation 7

  The correction coefficient α calculated by the correction coefficient calculation circuit 15 is input to the image correction circuit 16. The image correction circuit 16 performs luminance correction for each pixel on the image data read from the storage device 4 based on a correction function using the correction coefficient α. A correction function used for luminance correction in the luminance correction circuit 16 is expressed by the following equation (8).

  Output luminance value = correction coefficient α / input luminance value Equation 8

FIG. 5 is a graph showing the correction function shown in Equation 8 with the horizontal axis representing the input luminance value and the vertical axis representing the output luminance value (correction value). In the graph (brightness conversion straight line) shown in FIG. 5, the alternate long and short dash line indicates a correction function when the predicted noise amount of the face image portion after luminance conversion is equal to or less than the allowable noise amount (in the case of Equation 3), and the solid line indicates An example of a correction function when the predicted noise amount of the face image portion after luminance conversion exceeds the allowable noise amount (in the case of Expression 4) is shown respectively. ○ indicates the average luminance value (Yave) and the determined target average luminance value (the input target average luminance value (Ytarget) and the target luminance value when the predicted noise amount of the face image portion after luminance conversion is less than the allowable noise amount. The corresponding points with the same value) are the average luminance value (Yave) and the target luminance value to be determined when the predicted noise amount of the face image part after luminance conversion exceeds the allowable noise amount (Yave) The corresponding points with Ytarget ) are shown respectively. In the graph shown in FIG. 5, the luminance values (input luminance value and output luminance value) are represented by 256 levels from 0 to 255 by 8-bit data.

  As described above, in the image correction apparatus 1, the noise amount to be predicted when the luminance correction is performed on the image including the face image portion based on the input target average luminance value is the allowable noise amount. If it exceeds (Equation 4), a new target average luminance value (determined target average luminance value) is calculated so that the noise amount of the face image portion after correction becomes the same value as the allowable noise amount. be able to. Then, the input image data is corrected by a correction function according to the calculated determined target average luminance value. It is possible to reliably obtain a corrected face image portion in which noise is not conspicuous. Of course, when the noise of the image data representing the face image portion is small (in the case of Equation 3), luminance correction is realized in accordance with the user's request (whether the corrected face image portion should be brightened or darkened). . Luminance correction can be performed in consideration of both the user's request regarding noise amount and the user's request regarding luminance correction.

In the first embodiment described above, a numerical value (luminance value) is input as the target average luminance value (Yave). Of course, inputs such as “brighter” and “darker” are input from the input device 2. Also good. In this case, a relatively large value is set as the target average luminance value (Yave) corresponding to the “bright” input, and a relatively small value corresponding to the “dark” input. It is set as the target average luminance value (Yave). The target average luminance value (Yave) to be set is stored in advance in the memory of the target average luminance value determining circuit 14. Similarly, the above-mentioned allowable noise amount σ ² may be set to “large”, “small”, and the like. This is the same in the second embodiment described later.

  The correction coefficient α given to the image correction circuit 16 is calculated using image data (original image data) read from the storage device 4 in the above-described embodiment. Instead of the original image data, the correction coefficient α may be calculated based on reduced image data obtained by reducing the original image data. In this case, the image correction apparatus 1 is provided with a reduction circuit for reducing the original image data. The face area division circuit 11, the noise amount calculation circuit 12, the average luminance calculation circuit 13, and the correction coefficient calculation circuit 15 each perform processing based on the reduced image data. The image correction circuit 16 performs luminance correction on the original image data in accordance with a correction function defined by the correction coefficient α obtained using the reduced image data. This also applies to the calculation of the correction value (lookup table) obtained in the second embodiment described below.

  FIG. 6 shows the electrical configuration of the image correction apparatus 1 of the second embodiment. The image correction apparatus 1 according to the first embodiment shown in FIG. 2 is different from the image correction apparatus 1 in that a correction value calculation circuit (lookup table creation circuit) 25 is provided instead of the correction coefficient calculation circuit 15.

  In the first embodiment described above, luminance correction is performed on the input image data based on the correction function (Equation 8) (luminance conversion line) using the correction coefficient α calculated by the correction coefficient calculation circuit 15. (See FIG. 5). In the second embodiment, instead of the luminance correction based on the correction function (Equation 8) (luminance conversion line) using the correction coefficient α, the correspondence relationship of the output luminance values to all the input luminance values (luminance conversion) is performed by interpolation processing. Curve).

The correction value calculation circuit (lookup table creation circuit) 25 uses a correspondence relationship between a plurality of (for example, three) input luminance values and output luminance values obtained in advance, and handles other input luminance values. This circuit calculates an output luminance value by interpolation processing (creates a lookup table (LUT)). As described above, in the target average luminance determination circuit 14, the average luminance value (Yave) that is one of the input luminance values and the determined target average luminance value ( Ytarget ) that is the output luminance value corresponding to the average luminance value ( Yave ). Can be obtained. In the second embodiment, as a correspondence relationship between other input luminance values and output luminance values obtained in advance, a correspondence relationship in which the lowest luminance value is used as the lowest luminance value and the highest luminance value is used as is is the highest luminance value. . The minimum luminance value and the maximum luminance value may be the minimum and maximum luminance values that can be taken as the luminance value (for example, 0 and 255 in the case of 256 levels from 0 to 255), or for each pixel in the input image. Of these luminance values, the lowest and highest luminance values may be taken as the lowest and highest luminance values that can be taken as luminance values.

  In the correction value calculation circuit (lookup table creation circuit) 25, output luminance values (correction values) corresponding to input luminance values other than the above three input luminance values are calculated by interpolation processing. For the interpolation processing, a spline interpolation method, a nearest neighbor method, a bi-linear method, a bi-cubic method, or the like can be used. FIG. 7 shows a luminance conversion curve (gradation curve) obtained by a secondary spline interpolation method. The alternate long and short dash line indicates a luminance conversion curve based on a correction value (lookup table) obtained when the predicted noise amount of the face image portion after luminance correction is equal to or less than the allowable noise amount, and the solid line indicates a face image after luminance correction. A luminance conversion curve based on a correction value (lookup table) obtained when the predicted noise amount of the portion exceeds the allowable noise amount is shown.

1 is a block diagram illustrating an electrical configuration of a digital print system. It is a block diagram which shows the electric constitution of an image correction apparatus. (A) shows the image represented by the input image data, (B) shows the face image sectioned, and (C) shows the face area information. An example of the filter used for noise calculation processing is shown. A luminance conversion straight line is shown. It is a block diagram which shows the electrical structure of the image correction apparatus of 2nd Example. A luminance conversion curve is shown.

Explanation of symbols

1 Image correction device 2 Input device 3 Display device 4 Storage device
11 Face area division circuit
12 Noise amount calculation circuit
13 Average luminance value calculation circuit
14 Target average luminance value decision circuit
15 Correction coefficient calculation circuit
16 Image correction circuit
25 Correction value calculation circuit

Claims (8)

Representative luminance value calculating means for calculating a representative luminance value in a specific image included in an image represented by given image data;
A noise amount calculating means for calculating a noise amount included in the specific image data representing the specific image;
Target luminance value determining means for determining a target luminance value according to the noise amount calculated by the noise amount calculating means;
Means for creating data representing a luminance conversion line or a luminance conversion curve so that the representative luminance value calculated by the representative luminance value calculating means is corrected to the target luminance value determined by the target luminance value determining means; And luminance correction means for correcting the luminance of at least the specific image data representing the specific image based on the generated data representing the luminance conversion line or the luminance conversion curve,
An image correction apparatus comprising: A desired target luminance value input means for receiving an input of a desired target luminance value;
Based on the representative luminance value calculated by the representative luminance value calculating unit, the desired target luminance value input by the target luminance value input unit, and the noise amount calculated by the noise amount calculating unit, the specific image after luminance correction A prediction noise amount calculating means for calculating a prediction noise amount that will be included in the data;
The target brightness value determining means is:
The target luminance value is determined according to the predicted noise amount calculated by the predicted noise amount calculating means.
The image correction apparatus according to claim 1. The target brightness value determining means is:
Based on the allowable noise amount for noise that will be included in the specific image data after brightness correction, which is set or input in advance, and the calculated noise amount or the calculated predicted noise amount, The target luminance value is determined so that the noise amount is less than the allowable noise amount.
The image correction apparatus according to claim 1. The target brightness value determining means is:
If the calculated predicted noise amount is less than or equal to the allowable noise amount, the desired target luminance value input from the desired target luminance value input means is determined as the target luminance value,
If the calculated predicted noise amount exceeds the allowable noise amount, a new target luminance value that makes the predicted noise amount equal to the allowable noise amount is changed to the representative luminance value, allowable noise amount, and noise amount. And calculating the calculated new target luminance value as the target luminance value.
The image correction apparatus according to claim 3.   The image correction apparatus according to claim 1, wherein the specific image is a human face image.   The image correction apparatus according to claim 1, wherein the representative luminance value is an average luminance value in a specific image. Calculating a representative luminance value in a specific image included in an image represented by given image data;
Calculating the amount of noise included in the specific image data representing the specific image,
The target brightness value is determined according to the calculated noise amount,
Create data representing a luminance conversion line or luminance conversion curve so that the calculated representative luminance value is corrected to the determined target luminance value.
Based on the created luminance conversion line or data representing the luminance conversion curve, at least the specific image data representing the specific image is corrected for luminance,
Image correction method. A procedure for calculating a representative luminance value in a specific image included in an image represented by given image data;
A procedure for calculating the amount of noise included in the specific image data representing the specific image;
A procedure for determining a target luminance value according to the calculated amount of noise;
A procedure for creating data representing a luminance conversion line or a luminance conversion curve so that the calculated representative luminance value is corrected to the target luminance value determined by the target luminance value determining means;
A procedure for correcting the luminance of at least the specific image data representing the specific image based on the generated luminance conversion line or data representing the luminance conversion curve;
A program that causes a computer to execute.

Images