JP2009237657A - Image determination device, method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique for suppressing erroneous determination in image quality determination. <P>SOLUTION: A printer 10 includes an edge area specification unit 315 specifying an edge area of which the edge amount is larger than a reference value in an image: a frequency conversion unit 320 generating a power spectrum in which the edge area is expressed by a frequency area; a camera shake direction estimation unit 330 specifying a camera shake direction of an amplitude value in the power spectrum; a one-dimensional conversion unit 340 generating one-dimensional amplitude data in which the amplitude value of the camera shake direction is converted into one-dimensional form; a weighting unit 350 weighting the amplitude data according to a frequency; and a determination unit 360 determining whether or not the image is a camera shake image on the basis of the frequency in an area in which the amplitude value varies in a V shape in the amplitude data. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image determination technique for determining the characteristics of an image based on image data, and in particular, whether or not a photographic image based on image data is blurred due to camera shake, that is, whether or not it is “camera shake”. (Hereinafter, also referred to as “camera shake determination”).

  Pictures that have been shaken (hereinafter also referred to as “blurred images”) are less likely to be used for secondary use, such as printing and editing, compared to normal images that are not shaken. Various camera shake determination techniques have been proposed in order to distinguish a normal image and a camera shake image from among a plurality of image data in which images are recorded. For example, in Patent Document 1 below, an image for determining whether or not a defective image is blurred due to camera shake or defocusing based on the component ratio between a high-frequency component and a low-frequency component extracted separately from the image. A quality judgment technique is disclosed.

JP 2006-58749 A

  However, in the conventional image quality determination technique, an image photographed in the close-up (macro) mode or an image photographed with a relatively small F value has few in-focus portions and relatively few high-frequency components. There is a problem that even a normal image may be erroneously determined to be a defective image.

  In view of the above-described problems, an object of the present invention is to provide a technique capable of suppressing erroneous determination in image quality determination.

  SUMMARY An advantage of some aspects of the invention is to solve at least a part of the problems described above, and the invention can be implemented as the following forms or application examples.

  Application Example 1 An image determination apparatus according to Application Example 1 is an image determination apparatus that determines whether an image based on image data is blurred due to camera shake, and an edge having an edge amount larger than a reference value in the image. An edge region specifying unit for specifying a region, a frequency converting unit for generating a power spectrum representing an image of a spatial region in the edge region specified by the edge region specifying unit in a frequency region, and the frequency converting unit A camera shake direction estimator for identifying the second principal component direction of the amplitude value in the power spectrum, and a one-dimensional one-dimensional representation of the amplitude value in the region substantially along the second principal component direction in the power spectrum on the frequency axis A one-dimensional unit for generating amplitude data, and the one-dimensional amplitude data generated by the one-dimensional unit are weighted according to frequency. A weighting unit; and a determination unit that determines whether the image is blurred due to camera shake based on a frequency of a region in which an amplitude value changes in a valley shape in the one-dimensional amplitude data weighted by the weighting unit. It is characterized by that. Here, a camera shake image resulting from a linear movement of a hand or body is an image obtained by convolving a rectangular function with a subject in real space, and a sink function (sinc function) is convolved in a frequency space. It will be an image. Therefore, according to the image determination device of Application Example 1, it is possible to perform the camera shake determination based on the degree of influence of the sync function that appears in the camera shake direction in the frequency domain image. As a result, erroneous determination in camera shake determination can be suppressed even for an image with relatively few high-frequency components. In addition, since the hand shake determination can be performed by narrowing down to the edge region, it is possible to reduce the processing load for evaluating the degree of influence of the sync function appearing in the hand shake direction.

  Application Example 2 In the image determination apparatus according to Application Example 1, when the image data is image data that has been transform-encoded from a spatial domain to a frequency domain, the edge region specifying unit is configured to convert the transform code based on the image data. The edge region may be specified on the basis of an evaluation value calculated from the weighting factor. According to the image determination apparatus of Application Example 2, since the edge amount can be evaluated using the weighting coefficient of transform coding based on the image data, it is possible to reduce the processing load for specifying the edge region. .

  [Application Example 3] In the image determination apparatus according to Application Example 1 or 2, the frequency transforming unit is configured to perform either the discrete Fourier transform (DFT) or the fast Fourier transform (FFT). May be generated in the frequency domain. According to the image determination apparatus of Application Example 3, it is possible to obtain a frequency resolution sufficient to specify a region where the amplitude value changes in a valley shape due to the influence of the sync function.

  Application Example 4 In the image determination apparatus according to any one of Application Examples 1 to 3, the weighting unit multiplies the one-dimensional amplitude data generated by the one-dimensionalization unit by a frequency to thereby generate the one-dimensional amplitude data. When the frequency at which the amplitude value takes the minimum value in the frequency range of 50 Hz to 575 Hz in the one-dimensional amplitude data weighted by the weighting unit is 60 Hz to 300 Hz, It may be determined that the image is blurred due to camera shake. According to the image determination device of Application Example 4, the processing load required for the camera shake determination can be reduced by performing the camera shake determination by narrowing down to a frequency region that can be recognized by a human as a camera shake image.

  Application Example 5 In the image determination device according to any one of Application Examples 1 to 4, the edge region is specified by the edge region specifying unit, and the image is not blurred due to camera shake. When determined by the camera shake determination unit, an in-focus determination unit that determines that the image is in focus may be provided. According to the image determination apparatus of Application Example 5, in addition to specifying a camera shake image, it is possible to specify a focused image in focus.

  Application Example 6 In the image determination device according to any one of Application Examples 1 to 5, a defocus determination that determines that the image is not in focus when the edge region is not specified by the edge region specifying unit. A part may be provided. According to the image determination apparatus of the application example 6, in addition to specifying a camera shake image, an out-of-focus image that is out of focus can be specified.

  The image determination apparatus according to any one of Application Examples 1 to 6, further comprising: a printing unit that prints an image reproduced from the image data; and a determination unit that performs an operation by the determination unit among images reproduced from the image data. The image processing apparatus may further include a prohibition unit that prohibits printing of the image determined to be blurred by the printing unit. Thus, when an image reproduced from the image data is printed, it is possible to suppress a hand shake image from being printed.

  The image determination apparatus according to any one of Application Examples 1 to 6, further distinguishing between an image determined not to be camera shake by the determination unit and an image determined to be camera shake by the determination unit. You may provide the display part to display. As a result, when displaying an image to be reproduced from the image data, it is possible to display the image while distinguishing whether there is camera shake or not.

  The form of the present invention is not limited to the image processing apparatus, and other forms such as a printing apparatus (printer), an image display apparatus (viewer) incorporating the image processing apparatus, an image determination method using a computer, a computer program, and the like. It can also be applied to. Further, the present invention is not limited to the above-described embodiments, and it is needless to say that the present invention can be implemented in various forms without departing from the spirit of the present invention.

  In order to further clarify the configuration and operation of the present invention described above, an image processing apparatus to which the present invention is applied will be described below. In this embodiment, a printer having a function as an image processing apparatus will be described as an example of an image processing apparatus.

A. First embodiment:
A1. Printer configuration:
FIG. 1 is an explanatory diagram mainly showing the external configuration of the printer 10. The printer 10 is an ink jet printer that prints data such as characters and images by ejecting ink droplets onto a print medium 900 such as paper or a label. In this embodiment, the printer 10 has various functions such as a scanner and a copy as a so-called multifunction machine.

  The printer 10 includes a card slot 140 and a communication connector 150. The card slot 140 of the printer 10 is an interface that connects to a memory card 810 containing a storage medium such as a flash memory or a small hard disk so that data can be exchanged. The communication connector 150 of the printer 10 is an interface that connects to an external device 820 such as a personal computer, a digital still camera, or a digital video camera so as to exchange data. In this embodiment, the printer 10 has a function of printing image data stored in a memory card 810 connected to the card slot 140 or an external device 820 connected to the communication connector 150. In this embodiment, the printer 10 further performs image quality determination on the image data stored in the memory card 810 or the external device 820, thereby printing a non-blurred image or an unfocused image. It has a “printing function with image quality determination” that is selected as a target and excludes a camera shake image or a defocused image from a print target. Details of the “printing function with image quality determination” executed by the printer 10 will be described later.

  The printer 10 further includes a scanner unit 130, a display 160, and an operation panel 170. The scanner unit 130 of the printer 10 reads a document placed on a document table and converts it into digital data (scan). The display 160 of the printer 10 displays characters and images for the user of the printer 10. The operation panel 170 of the printer 10 receives an instruction input from the user of the printer 10.

  FIG. 2 is an explanatory diagram mainly showing the internal configuration of the printer 10. In addition to the card slot 140 and communication connector 150 described in FIG. 1, the printer 10 further includes a main control unit 110 that controls each unit of the printer 10 and a printing mechanism unit 120 that executes printing on the printing medium 900. Prepare.

  The main control unit 110 of the printer 10 includes an image reading unit 310, an edge region specifying unit 315, a frequency converting unit 320, and a camera shake direction estimating unit as functional elements for realizing the “printing function with image quality determination”. 330, a one-dimensionalization unit 340, a weighting unit 350, a determination unit 360, and a print instruction unit 370. The image reading unit 310 of the main control unit 110 reads image data from the memory card 810 connected to the card slot 140 or the external device 820 connected to the communication connector 150. The edge region specifying unit 315 of the main control unit 110 specifies an edge region in which the edge amount in the image based on the image data read by the image reading unit 310 is larger than the reference value. The frequency conversion unit 320 of the main control unit 110 generates a power spectrum in which an image of a spatial region in the edge region specified by the edge region specifying unit 315 is represented in the frequency domain. The camera shake direction estimation unit 330 of the main control unit 110 identifies the second principal component direction of the amplitude value in the power spectrum generated by the frequency conversion unit 320. The one-dimensionalization unit 340 of the main control unit 110 is a one-dimensional amplitude obtained by making the amplitude value in the region of the power spectrum substantially along the second principal component direction specified by the camera shake direction estimation unit 330 one-dimensionally on the frequency axis. Generate data. The weighting unit 350 of the main control unit 110 weights the one-dimensional amplitude data generated by the one-dimensionalization unit 340 according to the frequency. The determination unit 360 of the main control unit 110 determines whether the image based on the image data is blurred due to camera shake based on the frequency of the region where the amplitude value changes in a valley shape in the one-dimensional amplitude data weighted by the weighting unit 350. Determine whether. The print instruction unit 370 of the main control unit 110 instructs the print mechanism unit 120 to print a normal image that is determined not to be shaken by the determination unit 360.

  In this embodiment, the main control unit 110 of the printer 10 includes a central processing unit (hereinafter referred to as CPU), a read only memory (hereinafter referred to as ROM), a random access memory (Random Access Memory, ASIC (Application Specific Integrated Circuits) having hardware such as RAM) is provided. In the present embodiment, the main control unit 110 includes an image reading unit 310, an edge region specifying unit 315, a frequency converting unit 320, a camera shake direction estimating unit 330, a one-dimensionalizing unit 340, a weighting unit 350, a determining unit 360, printing. Software for realizing each function of the instruction unit 370 is installed. Details of the operation of the main controller 110 will be described later.

  As shown in FIG. 2, the printing mechanism unit 120 of the printer 10 includes a head unit 210, a carriage 200, a carriage drive unit 240, and a conveyance unit 250. The head unit 210 of the printing mechanism unit 120 has a total of six ejection heads 211, 212, 213, 214, 215, 216 for each color ink of black, cyan, light cyan, magenta, light magenta, and yellow. Each of the ejection heads 211 to 216 ejects ink by adjusting the voltage of a piezoelectric element (not shown). The carriage 200 of the printing mechanism unit 120 holds the head unit 210. The carriage drive unit 240 of the printing mechanism unit 120 drives the carriage 200 in the main scanning direction above the print medium 900. The transport unit 250 of the printing mechanism unit 120 transports the print medium 900 in the sub-scanning direction that intersects the main scanning direction in which the carriage 200 moves. Printing units 211 to 216 of the printing mechanism unit 120, the carriage driving unit 240, and the conveying unit 250 cooperate with each other based on instructions from the main control unit 110, thereby realizing printing on the printing medium 900.

A2. Printer behavior:
A2-1. Image printing process:
FIG. 3 is a flowchart illustrating image printing processing executed by the main control unit 110 of the printer 10. The image printing process in FIG. 3 is a process for realizing the above-described “printing function with image quality determination”. In the present embodiment, the execution request for executing the “print function with image quality determination” for the image data stored in the memory card 810 attached to the card slot 140 or the external device 820 attached to the communication connector 150. Is received from the user of the printer 10 via the operation panel 170, the main control unit 110 of the printer 10 starts the image printing process of FIG. In the present embodiment, the image printing process of FIG. 3 is realized by the CPU of the main control unit 110 operating based on software, but as another embodiment, the electronic circuit of the main control unit 110 is physically It may be realized by operating based on a simple circuit configuration.

  When starting the image printing process of FIG. 3, the main control unit 110 of the printer 10 executes a printing condition setting process for setting printing conditions (step S100). The print conditions set in the print condition setting process (step S100) include image data to be printed, print paper, print size, number of prints, print quality, and the like. In this embodiment, the printing conditions set in the printing condition setting process (step S100) are set based on information received from the user of the printer 10 via the operation panel 170. However, as another embodiment, It may be set based on preset printing conditions set in advance.

  After the print condition setting process (step S100), the main control unit 110 stores the image data stored in the memory card 810 or the external device 820 within the range of the print condition set in the print condition setting process (step S100). By selecting sequentially, image data to be determined for camera shake determination is prepared (step S302). In this embodiment, the image data to be determined is selected according to the order of the file names. However, as another embodiment, the image data is selected according to the order according to various attributes such as the file format, shooting date / time, and file update date / time. It may be.

  After the image data is prepared (step S302), the main control unit 110 executes an image quality determination process for determining whether an image based on the prepared image data is a camera shake image or a defocused image (step S302). Step S200). Details of the image quality determination process (step S200) will be described later.

  After the image quality determination process (step S200), the main control unit 110 determines whether the image data determined in the image quality determination process (step S200) is a camera shake image or a blurred image (step S304). ). When the image data determined in the image quality determination process (step S200) is a normal image (step S304), the main control unit 110 operates as the print instruction unit 370 to select the image as a print target. Then, the print mechanism unit 120 is instructed to print the image on the print medium 900 (step S306). On the other hand, when the image data determined in the image quality determination process (step S200) is a camera shake image or a blurred image (step S304), the main control unit 110 excludes the image from the print target, thereby Printing of the image by the printing mechanism unit 120 is prohibited (step S307).

  After printing execution or printing prohibition is processed for an image whose image quality is determined (steps S306 and S307), the main control unit 110 performs image quality determination processing until all image data to be subjected to image quality determination is processed. The processing from (Step S200) is repeatedly executed (Steps S308 and S309). After completing the processing for all the image data (step S308), the main control unit 110 executes a result display process for displaying the processing result of the image quality determination (step S200) to the user (step S500).

  FIG. 4 is an explanatory diagram illustrating an example of the result display screen 60 displayed on the display 160 in the result display process (step S500). The result display screen 60 of FIG. 4 includes an image list unit 610 that displays a list of image data determined in the image quality determination process (step S200). In the present embodiment, among the image data displayed in the image list unit 610, image data determined as a camera shake image or a blurred image in the image quality determination process (step S200), that is, image data for which printing is prohibited. Is attached with a thick frame 620, and a camera shake image, a defocused image, and a normal image are displayed separately.

A2-2. Image quality judgment processing:
FIG. 5 is a flowchart showing details of the image quality determination process (step S200) in FIG. When the main control unit 110 of the printer 10 starts the image quality determination process (step S200) of FIG. 5, the focus flag held in the memory (not shown) of the main control unit 110 is turned off (the binary number “0”). ”) (Step S205). Thereafter, the main control unit 110 operates as the image reading unit 310 to execute an image reading process of reading image data to be determined from the memory card 810 or the external device 820 (step S210). In the present embodiment, the image data read from the memory card 810 or the external device 820 in the image reading process (step S210) is JPEG data compressed in JPEG (JPEG) format, and a discrete cosine transform (Discrete Cosine Transform, DCT).

  After the image reading process (step S210), the main control unit 110 performs a determination area for performing image quality determination as a part of an area in the image based on the image data read in the image reading process (step S210). The determination area setting process for setting is performed (step S220).

  FIG. 6 is an explanatory diagram showing an example of how the determination area setting process (step S220) is performed. An image 50 in FIG. 6 schematically represents an image based on the image data read out in the image reading process (step S210) as raster data. In this embodiment, as shown in FIG. 6, the determination area 540 is defined as a rectangular area smaller than the image 50. In this embodiment, when the determination area setting process (step S <b> 220) is first performed on the image 50, the determination area 540 is set at the upper left corner of the image 50. When the determination area setting process (step S220) is performed on the image 50 for the second time or later, the determination area 540 is set to the right of the determination area 540 set the previous time and reaches the right end of the image 50. When set, the same setting is made from the left end one step down to the right. In this embodiment, the size of the determination area 540 is an area corresponding to 128 pixels × 128 pixels in the vertical and horizontal directions, but is appropriately determined according to the printing conditions such as the number of pixels of the image 50, the print size, and the print resolution. Can be set.

  After the determination area setting process (step S220), the main control unit 110 executes the edge area specifying process by operating as the edge area specifying unit 315 (step S500). In the edge area specifying process (step S500), the main controller 110 determines that the determination area 540 is an edge area when the edge amount of the determination area 540 set in the determination area setting process (step S220) is larger than the reference value. judge. Details of the edge region specifying process (step S500) will be described later.

  When the determination area 540 is determined to be an edge area in the edge area specifying process (step S500) (step S230), the main control unit 110 sets the focus flag to ON (binary number “1”). (Step S240). Thereafter, the main control unit 110 executes a camera shake determination process for determining whether an image based on the determination area 540 determined as the edge area in the edge area specifying process (step S500) is a camera shake image (step S100). S400). Details of the camera shake determination process (step S400) will be described later.

  When it is determined that the image based on the determination area 540 is a camera shake image in the camera shake determination process (step S400) (step S250), the main control unit 110 reads the image read out in the image reading process (step S210). After determining that the data is a camera shake image (step S286), the image quality determination process (step S200) is terminated.

  On the other hand, when the determination area 540 is not determined to be an edge area in the edge area specifying process (step S500) (step S230), or the image based on the determination area 540 is a camera shake image in the camera shake determination process (step S400). If it is determined that the determination area 540 is not determined (step S250), the main control unit 110 determines whether or not the determination area 540 is set in the entire image 50 based on the image data read in the image reading process (step S210). It is determined whether or not the region 540 has been set up to the lower right of the image 50 (step S260). When the determination region 540 is not set for the entire image 50 (step S260), the main control unit 110 repeatedly executes the processing from the determination region setting process (step S220).

  When the determination area 540 is set for the entire image 50 (step S260), the main control unit 110 determines whether or not the focus flag is on (binary number “1”) (step S270). ). When the focus flag is on (binary number “1”) (step S282), the main control unit 110 focuses the image data read in the image reading process (step S210) in focus. (Step S282), the image quality determination process (step S200) is terminated. On the other hand, when the focus flag is off (binary number “0”) (step S282), the main control unit 110 determines that the image data read in the image reading process (step S210) is out of focus. After determining that the image is out of focus (step S284), the image quality determination process (step S200) is terminated.

A2-3. Edge area identification processing:
FIG. 7 is a flowchart showing details of the edge region specifying process (step S500) in FIG. When the main control unit 110 of the printer 10 starts the edge region specifying process (step S500) in FIG. 7, the image data read out in the image reading process (step S210) is subjected to inverse Huffman coding and inverse quantization. By performing, a discrete cosine transform coefficient (DCT coefficient) in the determination area 540 set in the determination area setting process (step S220) is generated (step S510). Thereafter, the main control unit 110 calculates an edge amount E in the determination region 540 based on the DCT coefficient (step S520). In this embodiment, the edge amount E is calculated using a luminance change pattern associated with a DCT coefficient. When the edge amount E is greater than or equal to the threshold T (step S540), the main control unit 110 determines that the determination region 540 set in the determination region setting process (step S220) is an edge region (step S550). On the other hand, when the edge amount E is smaller than the threshold value T (step S540), the main control unit 110 determines that the determination region 540 set in the determination region setting process (step S220) is not an edge region (step S560). . The value of the threshold value T can be appropriately set according to printing conditions such as the number of pixels of the image 50, the print size, and the print resolution.

A2-4. Camera shake judgment processing:
8 and 9 are flowcharts showing details of the camera shake determination process (step S400) in FIG. When the camera shake determination process in FIG. 8 is started, the main control unit 110 of the printer 10 performs the frequency conversion process by operating as the frequency conversion unit 320 (step S420). In the frequency conversion process (step S420), the main control unit 110 generates a power spectrum in which the spatial domain image in the determination area 540 determined as the edge area in the edge area specifying process (step S500) is represented in the frequency domain.

  In the present embodiment, in the frequency conversion process (step S420), the main control unit 110 determines that the edge area is specified as the edge area in the edge area specifying process (step S500) among the JPEG data read in the image reading process (step S210). By decoding the portion of the data corresponding to the determination area 540, uncompressed raster data corresponding to the determination area 540 is prepared on the memory (not shown) of the main control unit 110 (step S422). The raster data means data representing an image by arranging a plurality of pixels (pixels) having color information vertically and horizontally.

  FIG. 10 is an explanatory diagram showing an example of how the frequency conversion process (step S420) is performed. A sample image 70 in FIG. 10 schematically represents raster data of a spatial region corresponding to the determination region 540 determined as the edge region in the edge region specifying process (step S500). In this embodiment, the size of the sample image 70 is 128 × 128 pixels in length and width, but may be appropriately set according to printing conditions such as print size and print quality in addition to the calculation capability of the main control unit 110. .

  Returning to the description of FIGS. 8 and 9, after the sample image 70 is prepared (step S422), the main control unit 110 generates a power spectrum by performing a fast Fourier transform on the sample image 70 (step S424). In this embodiment, the fast Fourier transform is used for generating the power spectrum. However, as another embodiment, discrete Fourier transform may be used.

  A power spectrum 80 in FIG. 10 schematically represents image data in the frequency domain generated by performing a fast Fourier transform on the sample image 70. In the power spectrum 80 of FIG. 10, by taking the horizontal frequency on the x-axis which is the horizontal axis and taking the vertical frequency on the y-axis which is the vertical axis, the absolute value of the complex coefficient after the fast Fourier transform is an amplitude value ( Scaled as (Power). The amplitude value in the power spectrum 80 indicates that the portion that appears white is a relatively large amplitude value, and the portion that appears black is a relatively small amplitude value. In the example shown in FIG. 10, since the sample image 70 is a camera shake image, the distribution of amplitude values in the power spectrum 80 shows a mode compressed in the camera shake direction from the second quadrant and the fourth quadrant toward the origin.

  Returning to the description of FIG. 8 and FIG. 9, after the frequency conversion process (step S420), the main control unit 110 executes the camera shake direction estimation process by operating as the camera shake direction estimation unit 330 (step S430). In the camera shake direction estimation process (step S430), the main control unit 110 specifies the second principal component direction of the amplitude value in the power spectrum 80 generated by the frequency conversion process (step S420).

  In the present embodiment, in the camera shake direction estimation process (step S430), the main control unit 110 converts the direct current component (DC component) indicating the average luminance of the image in the power spectrum 80 generated in the frequency conversion process (step S420). The dependent amplitude value is replaced with “0 (zero)” (step S432). Thereafter, the main control unit 110 multiplies the power spectrum 80 by a Hanning window function (Hanning Window). In this embodiment, the power spectrum 80 is processed by multiplying the Hanning window coefficient. However, as another embodiment, a window function such as a Hamming window function or a Blackman window function is used instead of the Hanning window function. Also good.

  FIG. 11 is an explanatory diagram showing an example of how the camera shake direction estimation process (step S430) is performed. The power spectrum 82 of FIG. 11 schematically represents image data in the frequency domain obtained by replacing the direct current component with “0” for the power spectrum 80 generated in the frequency conversion process (step S420) and then multiplying the Hanning window function. It is what. In the power spectrum 82, an x-axis which is a horizontal axis taking a horizontal frequency and a y-axis which is a vertical axis taking a vertical frequency are shown. In order to facilitate understanding of the drawing, the power spectrum 82 is compared. By expressing a part where a large amplitude value is distributed as a region 828, the distribution of amplitude values is shown in a simplified manner.

  Returning to the description of FIGS. 8 and 9, after the power spectrum 82 processed by the Hanning window function is generated (step S434), the main control unit 110 applies principal component analysis (PCA). The second principal component direction “y / x” obtained from the following formula 1 is estimated as the camera shake direction (step S436).

Here, λ is an eigenvector, and the coefficients V _xx , V _yy , and V _xy are obtained from the following Equation 2.

Here, the coefficient a _xy is a frequency amplitude value on the position (x, y) in the power spectrum 82. In the present embodiment, “−128 ≦ x, y ≦ 128” from the size of the sample image 70, and the coefficients a _x0 and a _0y depending on the DC component are “0”.

  The second principal component direction obtained from Equation 1 as “y / x” is a camera shake direction 822 that passes through the origins of the x-axis and y-axis in the power spectrum 82 and is orthogonal to the principal component direction 821, as shown in FIG. Identified as

  Returning to the description of FIGS. 8 and 9, after the camera shake direction estimation process (step S430), the main control unit 110 performs the one-dimensionalization process by operating as the one-dimensionalization unit 340 (step S440). In the one-dimensionalization process (step S440), the main control unit 110 sets the amplitude value in the region of the power spectrum 82 substantially along the camera shake direction 822 specified in the camera shake direction estimation process (step S430) on the frequency axis. One-dimensional one-dimensional amplitude data is generated.

  In the present embodiment, in the one-dimensionalization process (step S440), the main control unit 110 includes a central angle including the camera shake direction 822 estimated by the camera shake direction estimation process (step S430) among the regions in the power spectrum 82. A fan-shaped region having “” is specified (step S442). Thereafter, the main control unit 110 generates one-dimensional amplitude data by summing up the amplitude values in the fan-shaped region in the circumferential direction (step S444).

  FIG. 12 is an explanatory diagram illustrating an example of a state in which the one-dimensionalization process (step S440) is performed. The power spectrum 82 in FIG. 12 schematically represents image data in which the camera shake direction 822 is estimated by the camera shake direction estimation process (step S430). In the present embodiment, in the power spectrum 82 of FIG. 12, a fan-shaped region is set by dividing a 180-degree semicircular region over the first quadrant and the second quadrant into eight equal parts with the origins of the x-axis and y-axis as the center. Yes. Of these eight fan-shaped areas, the fan-shaped area 824 having a central angle including the camera shake direction 822 is specified by the main control unit 110 (step S422). In FIG. 12, one-dimensional amplitude data 84 in which the total value of the amplitude values is distributed on the one-dimensional frequency axis is generated by summing the amplitude values in the fan-shaped region 824 in the camera shake direction 822 in the circumferential direction. This is illustrated (step S444). In the one-dimensional amplitude data 84 of FIG. 12, frequency amplitude values are shown with the horizontal axis representing frequency and the vertical axis representing amplitude. The region extracted from the power spectrum 82 for the one-dimensional amplitude data 84 is not limited to the fan-shaped region obtained by dividing the semicircular region into eight equal parts, and the semicircular region is equally divided into eight or less or eight or more. It may be a fan-shaped area or a band-shaped area along the camera shake direction 822.

  Returning to the description of FIGS. 8 and 9, after the one-dimensionalization process (step S440), the main control unit 110 performs the weighting process by operating as the weighting unit 350 (step S450). In the weighting process (step S450), the main control unit 110 weights the one-dimensional amplitude data 84 generated in the one-dimensionalization process (step S440) according to the frequency.

  In the present embodiment, in the weighting process (step S450), the main control unit 110 smoothes the one-dimensional amplitude data 84 generated by the one-dimensionalization process (step S440) in an area including a frequency of 50 hertz (step). S452). The value of 50 Hz, which is the range for smoothing the one-dimensional amplitude data 84 in this embodiment, is set as a suitable value by the verification by the present inventor, but smoothes the one-dimensional amplitude data 84. The range is not limited to a frequency of 50 Hz, and an appropriate value may be appropriately set according to the image to be determined. In this embodiment, after the one-dimensional amplitude data 84 is smoothed (step S452), the main control unit 110 performs weighting by multiplying the smoothed one-dimensional amplitude data 84 by a frequency value ( Step S454).

  FIG. 13 is an explanatory diagram illustrating an example of a state in which the weighting process (step S450) is performed. In the one-dimensional amplitude data 84 of FIG. 13, frequency amplitude values smoothed in a region including a frequency of 50 Hz are shown with the horizontal axis representing frequency and the vertical axis representing amplitude. The one-dimensional amplitude data 86 in FIG. 13 shows frequency amplitude values weighted by multiplying frequencies with the horizontal axis representing frequency and the vertical axis representing amplitude. The frequency amplitude value of the one-dimensional amplitude data 86 is weighted as the frequency increases by weighting processing (step S450). For example, the frequency amplitude value of 60 Hz in the one-dimensional amplitude data 84 is multiplied by the frequency value “60”, and the frequency amplitude value of the frequency 300 Hz is multiplied by the frequency value “300”. Is multiplied.

  Returning to the description of FIG. 8 and FIG. 9, after the weighting process (step S450), the main control unit 110 executes the determination process by operating as the determination unit 360 (step S460). In the determination process (step S460), the main control unit 110 performs an image read process (step S450) based on the frequency of the region where the amplitude value changes in a valley shape in the one-dimensional amplitude data 86 generated by the weighting process (step S450). It is determined whether the image data read in S410 is a camera shake image.

  In the present embodiment, in the determination process (step S460), the main control unit 110 sets the frequency amplitude to a minimum value in the frequency range of 50 Hz to 575 Hz in the one-dimensional amplitude data 86 generated in the weighting process (step S450). The frequency fb to be taken is specified (step S462). In the present embodiment, the frequency range from 50 Hz to 575 Hz, which is the range for searching for the minimum value of the frequency amplitude, is set as a suitable value by verification by the present inventor. However, the minimum value of the frequency amplitude is searched. The range to be performed is not limited to a frequency of 50 Hz to 575 Hz, and an appropriate value may be appropriately set according to the image to be determined.

  When the frequency fb is larger than the frequency 60 hertz and smaller than the frequency 300 hertz, the main control unit 110 determines that the determination area 540 determined as the edge area in the edge area specifying process (step S500) is a camera shake image. (Step S466), the camera shake determination process (Step S400) is terminated. On the other hand, if the frequency fb is 60 Hz or less or 300 Hz or more, the main controller 110 determines that the determination region 540 determined as the edge region in the edge region specifying process (step S500) is normal with no camera shake image. After determining that the image is an image (step S468), the camera shake determination process (step S400) is terminated. In the present embodiment, the frequency range from 60 Hz to 300 Hz, which is the determination range of the frequency fb, is set as a suitable value by the verification by the present inventor, but the determination range of the frequency fb is from the frequency 60 Hz. The value is not limited to 300 Hz, and an appropriate value may be appropriately set according to the image to be determined. Note that the frequency 300 Hz, which is the determination range of the frequency fb in the present embodiment, corresponds to a value at which the camera shake width is about 0.3 mm when printing is performed with the L size (127 mm × 89 mm).

  FIG. 14 is an explanatory diagram showing an example of how the determination process (step S460) is performed. In FIG. 14, the one-dimensional amplitude data 86 generated by the weighting process (step S450) is shown with the frequency on the horizontal axis and the amplitude on the vertical axis. In the example shown in FIG. 14, the frequency fb at which the one-dimensional amplitude data 86 takes the minimum value pb in the frequency range of 50 Hz to 575 Hz is in the camera shake range that is greater than the frequency 60 Hz and smaller than 300 Hz. In this case, the main control unit 110 determines that the determination area 540 determined as the edge area in the edge area specifying process (step S500) is a camera shake image (step S466). In this embodiment, the frequency of the region where the amplitude value changes in a valley shape in the one-dimensional amplitude data 86 is specified as the frequency fb in which the amplitude value takes the minimum value pb. However, as another embodiment, the amplitude value is one-dimensional. You may specify as a frequency which takes the minimum value which satisfy | fills the conditions according to the weighting given to the amplitude data 86. FIG.

A3. Effect According to the printer 10 described above, a camera shake image resulting from a linear movement of a hand, a body, or the like becomes an image obtained by convolving a rectangular function (Equation 3) with respect to a subject in real space. Since it is considered that the power spectrum 80 is obtained by convolving the sync function (Equation 4) above, it is possible to perform camera shake determination based on the degree of influence of the sync function appearing in the camera shake direction in the frequency domain image. .

  Here, γ indicates a camera shake width, that is, a linear movement distance by camera shake, and k indicates a frequency.

  FIG. 15 is an explanatory diagram illustrating a relationship between a camera shake image and a sync function. In FIG. 15, the frequency amplitudes La, Lb, and Lc are shown with the horizontal axis representing frequency and the vertical axis representing amplitude. The frequency amplitude La indicates the frequency amplitude of a subject that is normally photographed without camera shake. The frequency amplitude Lb indicates the frequency amplitude of the sinc function expressed by the following Equation 4. The frequency amplitude Lb indicates the frequency amplitude of a camera shake image obtained by convolving the frequency amplitude Lb of the sync function with the frequency amplitude La of the subject. In FIG. 15, at points B1, B2, and B3 where the sink function of Equation 4 is “0”, the frequencies k are 1 / γ, 2 / γ, and 3 / γ.

  In the verification by the present inventor, the camera shake determination process (step S400) of the present embodiment was performed on 370 normal photographs and 110 camera shake photographs taken with a digital camera. As a result, 358 out of 370 normal photographs were obtained. It was determined that the image was a normal image (correct answer rate: 96.8%), and 69 out of 110 hand-blurred photos could be determined to be camera-shake images (correct answer rate: 62.7%). For these reasons, according to the printer 10, it is possible to suppress erroneous determination in camera shake determination even for an image with relatively few high-frequency components.

  In addition, since the camera shake determination process (step S400) can be performed by narrowing down to the determination area 540 that is an edge area, the processing load for evaluating the degree of influence of the sync function appearing in the camera shake direction 822 is reduced. Can be reduced. In addition, since the edge amount can be evaluated using the DCT coefficient of JPEG data, the processing load for specifying whether or not the determination area 540 is an edge area can be reduced.

Further, since the frequency conversion unit 320 of the printer 10 generates the power spectrum 80 by the fast Fourier transform (step S424), the frequency resolution sufficient to specify the region where the amplitude value changes in a valley shape due to the influence of the sync function. Can be obtained. In addition, since the camera shake direction estimation unit 330 of the printer 10 estimates the camera shake direction 822 from the power spectrum 82 multiplied by the Hanning window function (step S434), a specific error in the camera shake direction 822 can be suppressed. . Further, the camera shake direction estimation unit 330 of the printer 10 estimates the camera shake direction 822 by principal component analysis in which the amplitude values a _x0 and a _0y of the DC components in the power spectrum 80 are replaced with zero (step S432). By eliminating the influence of the component, the specific error in the camera shake direction 822 can be suppressed.

  In addition, the one-dimensionalization unit 340 of the printer 10 generates the one-dimensional amplitude data 84 from the fan-shaped region 824 in the camera shake direction 822 (step S442), so that the amplitude value in the direction close to the principal component direction is converted into the one-dimensional amplitude data. 84 can be excluded. In addition, the weighting unit 350 of the printer 10 smoothes the one-dimensional amplitude data 84 in a region including the frequency 50 Hz (step S452), so that the influence of the amplitude characteristic depending on the subject on the one-dimensional amplitude data 84 is suppressed. can do. Further, the weighting unit 350 of the printer 10 analyzes the one-dimensional amplitude data 86 in consideration of the degree of influence of the amplitude value according to the frequency in order to weight the one-dimensional amplitude data 84 by multiplying the frequency (step S454). be able to.

  In addition, the determination unit 360 of the printer 10 specifies the frequency fb in the frequency range of 50 to 575 hertz (step S462). Therefore, the processing load required for the camera shake determination is reduced by narrowing down the frequency range useful for the camera shake determination. Can be reduced. Further, the determination unit 360 of the printer 10 determines that the image is a camera shake image when the frequency fb is between 60 Hz and 300 Hz (step S466). The camera shake determination can be carried out.

A4. Variations:
In the above-described embodiment, the main control unit 110 of the printer 10 performs the edge region specifying process (step S500) that specifies the edge region based on the DCT coefficient. In other embodiments, the luminance value between pixels is determined. The edge region may be specified based on the difference. 16 and 17 are flowcharts showing the edge region specifying process (step S600) in the modification. FIG. 18 is an explanatory diagram illustrating a state in which edge extraction is performed based on a difference in luminance value between pixels. In the modification, in the image quality determination process (step S200), the main control unit 110 of the printer 10 replaces the edge area specifying process (step S500) in FIG. 7 with the edge area specifying process (step S500) in FIG. S600) is executed.

  When starting the edge area specifying process (step S600) in FIGS. 16 and 17, the main control unit 110 of the printer 10 calculates a difference ΔZ between the luminance values of pixels arranged in a predetermined direction in the determination area 540 (step S610). . In this modification, the luminance value difference ΔZ is calculated in both the horizontal direction and the vertical direction, but FIG. 18 shows how the luminance value difference ΔZ is calculated in the horizontal direction for the sake of simplicity. . FIG. 18A shows an example of the luminance value Z of each pixel, and FIG. 18B shows an example of the difference ΔZ of the luminance value Z in the case shown in FIG. 18A. . In this modification, the luminance value difference ΔZ is a positive and negative signed value. In the description of this modification example, the difference ΔZ is sequentially calculated between pixels. Therefore, if necessary, the difference of interest this time is expressed as ΔZ (n), and the difference between the pixels immediately before the pixel of interest is expressed. The difference is expressed as ΔZ (n−1).

  After the luminance value difference ΔZ is calculated (step S610), the main control unit 110 calculates and calculates the sign of the luminance value difference ΔZ (n) calculated for the pixel of interest for the previous pixel. It is determined whether or not it is the same as the sign of the difference ΔZ (n−1) in luminance value (step S612). When the sign of the brightness value difference ΔZ (n) matches the sign of the brightness value difference ΔZ (n−1) (step S612: “YES”), the main control unit 110 determines that the edge continues. Determination is made, and a cumulative value SZ obtained by accumulating the difference ΔZ (n) in luminance value is calculated (step S614). Specifically, the current difference ΔZ (n) is added to the accumulated value SZ that is reset to 0 when it is determined that the edge has started (SZ ← SZ + ΔZ (n)). After the difference ΔZ (n) is added to the accumulated value SZ (step S614), the main control unit 110 advances the position of the pixel of interest by 1 in a predetermined direction (step S616: n ← n + 1) and continues the above processing. (Steps S610 to S616).

  While repeating the above process, the sign of the difference ΔZ (n) in the luminance value calculated for the pixel of interest is the sign of the difference ΔZ (n−1) in the calculated luminance value calculated for the previous pixel. If it is determined that they are not the same (step S612: “NO”), the main control unit 110 determines whether or not the absolute value | SZ | of the accumulated value SZ obtained so far is greater than a predetermined threshold Th1. Judgment is made (step S620). FIG. 18B shows how this determination is made. When the absolute value of the cumulative value SZ is larger than the threshold value Th1 (step S620: “YES”), since there is a possibility that there is an edge to be determined, the main control unit 110 calculates the edge width We ( Step S622). The edge width We is set as the number of pixels in which the luminance value difference ΔZ continues with the same sign.

  After the edge width We is calculated (step S622), the main control unit 110 determines whether the edge width We is equal to or smaller than a predetermined threshold Th2 (step S624). When the edge width We is equal to or smaller than the threshold Th2 (step S624: “YES”), the main control unit 110 adds the value “1” to the edge amount E, assuming that these pixels are in focus ( Step S626). In the present modification, the edge amount E is reset to “0” at the start of the edge region specifying process (step S600). After the edge amount E is added (step S626), the main control unit 110 resets the cumulative value SZ to the value “0” (step S634), and advances the position of the pixel of interest by 1 in a predetermined direction (step S636). : N ← n + 1). On the other hand, if the determination in step S620 or S624 is “NO”, it is determined that the edge is not an in-focus state, and no process is performed, and the process proceeds to a process of resetting the accumulated value SZ in step S634. To do. As a result of these processes, as shown in FIG. 18, when the signs of the difference ΔZ in luminance values of adjacent pixels are not the same between adjacent pixels, the presence / absence of an edge is determined based on the cumulative value SZ of the difference ΔZ. Regardless of the determination, the accumulated value SZ is reset to 0 in preparation for the next determination.

  Therefore, when the determination for one edge is completed through the above-described series of processing, it is determined whether all the pixels have been determined (step S638). If not, the processing returns to step S610 to perform the above processing. The processing of (Steps S610 to S626) is repeated. If the above determination is completed for all the pixels (step S638: “YES”), the main control unit 110 determines whether or not the edge amount E is equal to or greater than the threshold T (step S640). When the edge amount E is greater than or equal to the threshold T (step S640), the main control unit 110 determines that the determination region 540 set in the determination region setting process (step S220) is an edge region (step S650). On the other hand, when the edge amount E is smaller than the threshold value T (step S640), the main control unit 110 determines that the determination area 540 set in the determination area setting process (step S220) is not an edge area (step S660). . The value of the threshold value T can be appropriately set according to printing conditions such as the number of pixels of the image 50, the print size, and the print resolution.

  According to the modification described above, the edge is extracted by determining the magnitude of the accumulated value SZ of the difference ΔZ when the sign of the luminance value difference ΔZ of the adjacent pixels is not the same between the adjacent pixels. It is done by. Therefore, when the arrangement of pixels constitutes an edge and there is an inter-pixel having a small difference ΔZ in the middle, the front and rear are not determined as separate edges. On the other hand, as an overall tendency, even if the luminance value of the pixel gradually increases or decreases, if the difference ΔZ becomes 0 or changes its sign in the middle, it is determined that the edge is divided there. The possibility of mistaking the edge existence area is reduced. As a result, even if there is noise or the like, it is possible to accurately determine the presence of an edge that should be handled as focusing information.

B. Other embodiments:
As mentioned above, although embodiment of this invention was described, this invention is not limited to such embodiment at all, Of course, it can implement with various forms within the range which does not deviate from the meaning of this invention. is there. For example, in the present embodiment, the edge region is specified using DCT coefficients based on JPEG data (step S500), but as another embodiment, transform codes using orthonormal transform such as Hadamard transform and discrete Fourier transform are used. In the case of the image data, the edge area may be specified by using the weighting coefficient of the transform coding, or the edge area may be specified by calculating the luminance change amount from the raster data. Alternatively, the edge region may be specified using a reduced image (thumbnail) generated by thinning out data from the image data. In this embodiment, the camera shake determination process (step S240) is performed every time the determination area 540 is identified as an edge area. However, as another embodiment, all the determination areas 540 are edge areas. The camera shake determination process (step S240) may be performed after specifying whether or not there is.

  In the above-described embodiment, the example in which the present invention is applied to a printer has been described. However, as another embodiment, the present invention is applied to an electronic device that handles image data, such as a personal computer, a digital camera, and a viewer. Also good.

2 is an explanatory diagram mainly illustrating an external configuration of a printer. FIG. 2 is an explanatory diagram mainly showing an internal configuration of a printer. FIG. 4 is a flowchart illustrating image printing processing executed by a main control unit 110 of the printer 10. It is explanatory drawing which shows an example of the result display screen 60 displayed on the display 160 by a result display process (step S500). It is a flowchart which shows the detail of the image quality determination process (step S200) in FIG. It is explanatory drawing which shows an example of a mode that the determination area | region setting process (step S220) is implemented. It is a flowchart which shows the detail of the edge area | region identification process (step S500) in FIG. It is a flowchart which shows the detail of the camera-shake determination process (step S400) in FIG. It is a flowchart which shows the detail of the camera-shake determination process (step S400) in FIG. It is explanatory drawing which shows an example of a mode that a frequency conversion process (step S420) is implemented. It is explanatory drawing which shows an example of a mode that camera shake direction estimation processing (step S430) is implemented. It is explanatory drawing which shows an example of a mode that a one-dimensionalization process (step S440) is implemented. It is explanatory drawing which shows an example of a mode that a weighting process (step S450) is implemented. It is explanatory drawing which shows an example of a mode that a determination process (step S460) is implemented. It is explanatory drawing which shows the relationship between a camera shake image and a sync function. It is a flowchart which shows the edge area | region identification process (step S600) in a modification. It is a flowchart which shows the edge area | region identification process (step S600) in a modification. It is explanatory drawing which shows a mode that edge extraction is performed from the difference of the luminance value between pixels.

Explanation of symbols

10 ... Printer 50 ... Image 60 ... Result display screen 70 ... Sample image 80 ... Power spectrum 82 ... Power spectrum 84 ... One-dimensional amplitude data 86 ... One-dimensional amplitude Data 110 ... Main control unit 120 ... Printing mechanism unit 130 ... Scanner unit 140 ... Card slot 150 ... Communication connector 160 ... Display 170 ... Operation panel 200 ... Carriage 210 ... head unit 211 ... jet head 240 ... carriage drive unit 250 ... conveying unit 310 ... image reading unit 315 ... edge region specifying unit 320 ... frequency converting unit 330 ... Camera shake direction estimation unit 340 ... One-dimensionalization unit 350 ... Weighting unit 360 ... Determination unit 370 ... Print instruction unit 540 ... Determination region 610 ... Image list unit 620 ... Thick Frame 810 ... Memory card 820 ... External device 821 ... Main component direction 822 ... Hand Blur direction 824 ... Fan-shaped area 828 ... Area 900 ... Printing medium fb ... Frequency pb ... Minimum value

Claims (8)

An image determination apparatus that determines whether an image based on image data is blurred due to camera shake,
An edge region specifying unit for specifying an edge region having an edge amount in the image larger than a reference value;
A frequency conversion unit that generates a power spectrum representing an image of a spatial region in the edge region specified by the edge region specifying unit in a frequency domain;
A camera shake direction estimation unit that identifies the second principal component direction of the amplitude value in the power spectrum generated by the frequency conversion unit;
A one-dimensionalization unit that generates one-dimensional amplitude data in which the amplitude value in a region substantially along the second principal component direction in the power spectrum is one-dimensionally plotted on the frequency axis;
A weighting unit that weights the one-dimensional amplitude data generated by the one-dimensionalization unit according to a frequency;
An image determination comprising: a camera shake determination unit that determines whether the image is blurred due to camera shake based on a frequency of a region where an amplitude value changes in a valley shape in the one-dimensional amplitude data weighted by the weighting unit. apparatus.   When the image data is image data that has been transform-encoded from a spatial domain to a frequency domain, the edge region specifying unit is based on an evaluation value calculated from a weight coefficient of the transform encoding based on the image data. The image determination apparatus according to claim 1, wherein the image determination apparatus specifies an edge region.   The image determination device according to claim 1, wherein the frequency conversion unit generates a power spectrum representing the image in a frequency domain by one of discrete Fourier transform and fast Fourier transform. The image determination device according to any one of claims 1 to 3,
The weighting unit weights the one-dimensional amplitude data by multiplying the one-dimensional amplitude data generated by the one-dimensionalization unit by a frequency,
When the frequency at which the amplitude value has a minimum value in the frequency range of 50 Hz to 575 Hz in the one-dimensional amplitude data weighted by the weighting unit is 60 Hz to 300 Hz, An image determination apparatus that determines that the image is blurred due to blurring.   Further, when the edge region is specified by the edge region specifying unit, and the image blur determining unit determines that the image is not blurred due to camera shake, the image is in focus. The image determination apparatus according to claim 1, further comprising an in-focus determination unit.   The image determination apparatus according to claim 1, further comprising a defocus determination unit that determines that the image is out of focus when the edge region is not specified by the edge region specification unit. An image determination method for determining whether an image based on image data is blurred due to camera shake using a computer,
An edge region specifying means provided in the computer specifies an edge region having an edge amount in the image larger than a reference value;
The frequency conversion means provided in the computer generates a power spectrum representing the spatial domain image in the edge area specified by the edge area specifying means in the frequency domain, and the camera shake direction estimation means provided in the computer includes the frequency conversion Identifying a second principal component direction of the amplitude value in the power spectrum generated by the means;
One-dimensionalization means provided in the computer generates one-dimensional amplitude data in which the amplitude value in a region substantially along the second principal component direction in the power spectrum is one-dimensionalized on the frequency axis;
A weighting means provided in the computer weights the one-dimensional amplitude data generated by the one-dimensionalization means according to the frequency;
A step of determining whether or not the image is blurred due to camera shake based on a frequency of a region where the amplitude value changes in a valley shape in the one-dimensional amplitude data weighted by the weighting unit; An image determination method comprising: A program for causing a computer to realize a function of determining whether an image based on image data is blurred due to camera shake,
An edge area specifying function for specifying an edge area having an edge amount in the image larger than a reference value;
A frequency conversion function for generating a power spectrum representing a spatial domain image in the edge area specified by the edge area specifying function in a frequency domain; and a second principal component direction of an amplitude value in the power spectrum generated by the frequency conversion function. Camera shake direction estimation function to identify
A one-dimensionalization function for generating one-dimensional amplitude data in which the amplitude value in a region substantially along the second principal component direction in the power spectrum is one-dimensional on the frequency axis;
A weighting function for weighting the one-dimensional amplitude data generated by the one-dimensionalization function according to the frequency;
In order to cause a computer to realize a determination function for determining whether or not the image is blurred due to camera shake based on the frequency of a region where the amplitude value changes in a valley shape in the one-dimensional amplitude data weighted by the weighting function Program.

Images