JP2014048260A - Inspection device, and inspection method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an inspection device that eliminates excessive inspection techniques while maintaining necessary quality, and an inspection method.SOLUTION: An inspection device evaluating quality of an evaluation image includes: image acquisition means for acquiring the evaluation image; contour extraction means for extracting contour data indicating contours included in the acquired evaluation image; and quantization means for analyzing a frequency of the extracted contour data, and quantizing the feature of the contour data on the basis of a spectrum value in a frequency band where sensitivity of human visual performance is high.

Description

  The present invention relates to an inspection apparatus for inspecting image quality.

  Conventionally, the following methods are known as methods for inspecting image quality. In the invention disclosed in Patent Document 1, features of an original image are collated by pattern matching, and a score for the collation result is searched using a dictionary. And quality evaluation is performed using this score. (For example, refer to Patent Document 1).

JP 2003-145898 A

  In the case of using the pattern matching method described above, it is necessary to extract the features of the target original image with high accuracy in order to increase the matching accuracy. For this reason, in the conventional inspection apparatus, fine feature components that do not affect human vision are extracted with high accuracy, and these feature components are applied to matching. As a result, the apparatus and processing required for the inspection are excessive.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide an inspection apparatus and an inspection method that eliminates an excessive inspection method while maintaining necessary quality.

  In order to solve the above problems, in the present invention, an inspection apparatus for evaluating the quality of an evaluation image, the image acquisition means for acquiring the evaluation image, and the contour included in the acquired evaluation image Contour extraction means for extracting the contour data to be shown, and quantification for analyzing the frequency of the extracted contour data and quantifying the characteristics of the contour data based on a spectrum value in a frequency band in which the sensitivity of human visual characteristics is high Means.

In the invention configured as described above, the contour extracting unit extracts the contour included in the evaluation image acquired by the image acquiring unit as the contour data. Then, the quantification means performs frequency analysis on the extracted contour data, and quantifies the characteristics of the contour data based on a spectrum value in a frequency band in which the sensitivity of human visual characteristics is high.
Therefore, contour features can be quantified based on frequency components specialized in human visual characteristics. That is, since it is not necessary to extract fine features that do not affect human visual characteristics, it is possible to evaluate image quality by simple processing while taking human vision into consideration.

Further, the quantification unit may be configured to indicate a relative positional relationship of each line segment forming the contour data by a predetermined periodic function and perform the frequency analysis on the periodic function.
In the invention configured as described above, the relative positional relationship of each component of the contour data can be treated as a feature, and rattling and unevenness included in the contour can be evaluated.

The quantified feature of the contour data may be acquired based on a deviation of a spectrum value in the frequency component.
In the invention configured as described above, the quantification can be performed more easily by using the deviation of the spectrum value.

Further, the contour extracting unit multi-values the evaluation image to extract the contour data, and the threshold value used in the multi-value conversion is set in consideration of human visual characteristics. Good.
In the invention configured as described above, the contour data obtained can be adapted to human perception by reflecting the human visual characteristics in the extraction of the contour data.

Moreover, it is good also as a structure which has the index value calculation means which calculates the index value which shows the quality of the said image for evaluation based on the characteristic of the said quantified outline data.
In the invention configured as described above, since the quality of the evaluation image can be shown as a numerical value, a more objective evaluation can be performed on the image.

And the said index value calculation means is good also as a structure which calculates the said index value based on the characteristic of the said quantified outline data, and the density | concentration characteristic of the part enclosed by the said outline.
In the invention configured as described above, the index value can be calculated according to the density characteristic of the image by using the density characteristic of the portion surrounded by the outline in addition to the feature of the outline data.

  Furthermore, the present invention can also be applied to an inspection method for inspecting an evaluation image.

It is a figure explaining the inspection method of the present invention. It is a block block diagram explaining the inspection apparatus 1 of this invention. 4 is a flowchart for explaining an inspection method executed by the inspection apparatus 1. It is a figure which shows the image for evaluation used by the test | inspection apparatus 1. FIG. It is a figure explaining the structure of the outline extraction part. It is a flowchart which shows the process of step S2 in detail. It is a figure explaining extraction of outline data. 4 is a block configuration diagram showing a more detailed configuration of a feature acquisition unit 43. FIG. It is a flowchart which shows the process of step S3 in detail. It is a figure explaining acquisition of the feature of outline data. It is a wave form diagram which shows a spectrum change. It is a flowchart which shows the process of step S5 in detail. It is a figure which shows the subjective value used in the case of calculation of an index value.

  Hereinafter, embodiments of the present invention will be described in the following order.

(1) Outline of Inspection Apparatus FIG. 1 is a diagram for explaining an inspection method of the present invention. FIG. 2 is a block diagram illustrating the inspection apparatus 1 of the present invention. FIG. 3 is a flowchart for explaining an inspection method executed by the inspection apparatus 1. Further, FIG. 4 is a diagram showing an evaluation image used by the inspection apparatus 1.

  As shown in FIG. 2, the inspection apparatus 1 includes a printer 10, a scanner 20, and a personal computer (PC) 30. In the inspection apparatus 1, the quality inspection by the PC 30 is performed using the evaluation image printed by the printer 10.

  A known printer 10 can be used, and may be an ink jet printer or a laser printer. In the present embodiment, the printer 10 prints an evaluation image using the evaluation data output from the PC 30.

  The evaluation image is an image used for the PC 30 to calculate the index value Ind_q. In the present embodiment, as shown in FIG. 4, an evaluation image including a character string recorded with six different character sizes (40 pt, 20 pt, 11 pt, 10.5 pt, 8 pt, 4 pt) is used. Each character string included in the evaluation image is referred to as a target image group. Each character included in the target image group is also referred to as a target image. In addition, what constitutes the image for evaluation is not limited to characters, and may be appropriately selected according to the inspection application.

  The scanner 20 reads the evaluation image printed by the printer 10 and outputs it to the PC 30. The evaluation image read by the scanner 20 is composed of pixels having a predetermined resolution (for example, 1200 dpi), and each pixel includes each color component value in the sRGB color space. A known scanner 20 can be used.

The PC 30 calculates an index value Ind_q for evaluating the evaluation image output from the scanner 20.
The PC 30 is an existing personal computer, and is connected to the printer 10 and the scanner 20. As shown in FIG. 2, the PC 30 includes a controller 31, IFs 32 and 33, a keyboard 34, an external memory 35, and a display 36.

  The PC 30 is connected to the printer 10 and the scanner 20 via IFs 32 and 33, respectively. The IFs 32 and 33 are connected to the controller 31.

The controller 31 includes a CPU, a ROM, and a RAM. The CPU 31 executes the program recorded in the ROM, so that the controller 31 has a color conversion unit (image acquisition unit) 41, a contour extraction unit (contour extraction unit) 42, a feature acquisition unit (quantification unit) 43, An index value calculation unit (index value calculation means) 44 is functionally provided.
Furthermore, the controller 31 can control driving of the printer 10 and the scanner 20 by executing a driver program (not shown) recorded in the external memory 35.

  The external memory 35 is configured by a large capacity memory such as an HDD or an SSD. The external memory 35 stores a program for the PC 20 to control the printer 10 or the scanner 20 and various data and programs used to calculate the index value Ind_q.

As shown in FIGS. 1 and 3, the following method is used to calculate the index value Ind_q executed by the PC 30.
First, in step S1 of FIG. 3, an evaluation image is acquired (FIG. 1A). In addition, the color conversion unit 41 converts the color components of the evaluation image. The color conversion unit 41 converts the color component values of the pixels constituting the evaluation image into values in another color space. Specifically, the color conversion unit 41 converts the color component value (sRGB color space) of the pixel of the evaluation image into an L * a * b * value (Lab color space). The conversion of color component values by the color conversion unit 41 is performed using a known conversion formula. Here, the conversion of the color component value of the pixel into a value in the Lab color space by the color conversion unit 41 is an example. For example, the value is converted into a value (X, Y, Z) in the XYZ color space. There may be.

  In step S2, the contour data is extracted by the contour extraction unit 42 (FIG. 1B). The contour data is data composed of pixels that define the boundary of the target image. In the present embodiment, the contour extraction unit 42 performs extraction by acquiring the coordinates (x, y) of the pixels constituting the contour data.

  In step S3, the feature acquisition unit 43 acquires the features of the contour data (FIG. 1 (c)). In the present embodiment, the feature of the contour data is acquired from the relative positional relationship between the components constituting the contour data. A case is assumed in which the curve represented by the contour data is defined as a contour line C, pixels (points) are connected, and the contour line C is approximated by a broken line. When an arbitrary line segment of the contour line C obtained by connecting pixels is linear in relation to the proximity component, the line segment can be regarded as a component with low backlash and unevenness. On the other hand, when an arbitrary line segment is not linear due to the proximity component, the line segment can be regarded as a component with high rattling and unevenness. That is, it is possible to determine that the contour line C including many components having high rattling and unevenness as described above has a high degree of shading.

  Therefore, the feature acquisition unit 43 expresses the relative positional relationship of each component included in the contour data as a periodic function w, and acquires a feature by performing frequency analysis on the periodic function w. This periodic function w expresses the contour line C expressed by the contour data as a function having the peripheral length as a cycle.

  Further, as shown in FIGS. 1D and 1E, the feature acquisition unit 43 extracts only the spectrum of the specific frequency component from the contour data features, and quantifies the contour data features based on this spectrum. (Rag amount Rag described later). The feature acquisition unit 43 quantifies the feature based on the obtained spectrum deviation.

  And in step S4, it implements until it quantifies with respect to all the target images used as the object of quantification (step S4: YES). For example, when evaluating the character size of 20pt, 10.5pt, and 4pt constituting the evaluation image, the features of the contour data are quantified for all these target images.

  Further, in step S5, the index value calculation unit 44 calculates the index value Ind_q (FIG. 1 (f)). This index value Ind_q is a value indicating the evaluation value of the evaluation image, and is calculated based on the features of the quantified contour data. Therefore, the evaluation image can be evaluated by using the index value Ind_q. Further, by evaluating the evaluation image using the index value Ind_q, a printer that prints an evaluation image with a high evaluation can be regarded as having high performance.

  Below, the Example as an example of the inspection apparatus 1 is described.

(2) Extraction of contour data First, the contour data extraction method executed in step S2 of FIG. 3 will be described. FIG. 5 is a diagram illustrating the configuration of the contour extraction unit 42. FIG. 6 is a flowchart showing in detail the process of step S2. FIG. 7 is a diagram for explaining extraction of contour data.

  As illustrated in FIG. 5, the contour extraction unit 42 includes a DFTC (Discrete Fourier Transform Converter) 421, a VTF (Visual Transfer Function) application unit 422, an IDFTC (Inverse Discrete Fourier Transform Converter) 423, and a binarization processing unit. 424 and a coordinate extraction unit 425.

In step S21 of FIG. 6, the DFTC 421 converts the pixels constituting the evaluation image into the spatial frequency characteristics F (u, v) (hereinafter, F (u, v) is also described as a spatial frequency component). The DFTC 421 calculates the spatial frequency characteristic F (u, v) using the discrete Fourier transform represented by the following formula (1).

  In step S22, the VTF application unit 422 weights each spatial frequency component according to visual characteristics. It is known that human visual sensitivity changes depending on spatial frequency. When this is graphed as a visual characteristic function VTF (u, v), as shown in FIG. 7A, the human visual sensitivity is highest in a predetermined low frequency band. The VTF application unit 422 weights the spatial frequency components of the pixels according to visual characteristics. The weight given by the VTF application unit 422 is set to a value between 0 and 1 by the visual characteristic function VTF (u, v). Therefore, for example, as shown in FIG. 7A, a high weight is assigned to a frequency component with high visual sensitivity. Note that the visual sensitivity is represented by using the visual characteristic function VTF (u, v). For example, a function approximating visual characteristics (for example, contrast sensitivity function CSF (Contrast Sensitivity Function)) may be used. Further, the visual characteristics may be used by applying filters to the a * and b * channels as well as the L * channel. Furthermore, a different filter may be applied to each channel of L * a * b *.

In step S23, the IDFTC 423 performs inverse Fourier transform on the spatial frequency component weighted in step S22 and converts it to a value in real space. For example, the IDFTC 423 converts a spatial frequency component into a value in real space using an inverse discrete Fourier transform represented by the following formula (2).

ここで、
Ｒｘは、閾値
Ｌ*ｍａｘは、明度の最大値
Ｌ*ｍｉｎは、明度の最小値
ｘは、視覚特性に応じて設定される値。 In step S24, the binarization processing unit 424 performs binarization processing on each pixel. By this binarization processing, the evaluation image is converted into a binary image in which the pixels having the lightness L * of the threshold value Rx or more are “1” and the pixels having the lightness L * of the threshold value Rx or less are “0”. FIG. 7B shows a threshold Rx setting method used for the binarization process as an example. In the present embodiment, the variation threshold method is used as a method for setting the threshold Rx used in the binarization process. Therefore, the threshold value Rx is set based on the following formula (3).

here,
Rx is a threshold value L * max, a lightness maximum value L * min, and a lightness minimum value x is a value set according to visual characteristics.

  In this embodiment, x is set to 60% to 70%. By reflecting the human visual characteristics in the setting of the threshold value Rx, the obtained contour data can be adapted to human perception. That is, it is possible to remove components that do not affect the vision.

  In step S25, the coordinate extraction unit 425 extracts contour data components (pixels) from the binarized image. Specifically, the coordinate extraction unit 425 first labels each target image included in the binarized image, and separates the target image from the evaluation image. Next, the contour extraction unit 42 extracts contour data by applying a contour extraction algorithm to the labeled target image.

  In this contour extraction algorithm, first, a predetermined target image is horizontally scanned to determine a start point (pixel) for performing contour search. Next, the vicinity centering on this starting point is searched counterclockwise to search for unsearched pixels. In this embodiment, the vicinity of the start point is searched for. Then, the first searched pixel is set as the next start point. The coordinates of the searched pixel are recorded as the contour data component. Further, the above process is repeated for all pixels. Therefore, the coordinate extraction unit 425 can acquire the coordinates (x, y) of the pixel that is the component of the contour data.

(3) Acquisition of Outline Data Features Next, an outline data feature acquisition method executed in step S3 of FIG. 3 will be described. FIG. 8 is a block configuration diagram showing a more detailed configuration of the feature acquisition unit 43. FIG. 9 is a flowchart showing in detail the process in step S3. FIG. 10 is a diagram for explaining the acquisition of the characteristics of the contour data.

  As shown in FIG. 8, the feature acquisition unit 43 includes a complex plane expansion unit 431, a periodic function calculation unit 432, a DFTC 433, a bandpass filter 434, an IDFTC 435, and a quantification unit 436.

  In step S31 of FIG. 9, the complex plane development unit 431 converts the coordinates (x, y) of each component (pixel) of the contour data into coordinates z on the complex plane. The complex plane is a coordinate space defined by a real part (x) and an imaginary part (yi). The complex plane development unit 431 converts x into a coordinate on the complex plane with x being a real part (x) and y being an imaginary part (yi) among the coordinates (x, y) of the components of the contour data.

δはｚ（ｊ）とｚ（ｊ＋１）との間の距離でありδ＝｜ｚ（ｊ＋１）−ｚ（ｊ）｜により算出される。ここで、上記式（４）を用いて得られる周期関数をＰ表現とも記載する。また、Ｐ表現をフーリエ変換して得られるものをＰ型フーリエ記述子ｃ（ｋ）とも記載する。 In step S <b> 32, the periodic function calculation unit 432 acquires a periodic function w indicating characteristics for each of the real part (x) and the imaginary part (yi) of the coordinate z on the complex plane. As shown in FIG. 10 (a), the relationship between the points z (j) and z (j + 1) constituting the curve obtained by approximating the contour line C by a polygonal line is a complex number by using the argument θ (j). It can be expressed using a periodic function w (j) which is a value function. The periodic function w (j) is expressed by the following equation (4).

δ is a distance between z (j) and z (j + 1), and is calculated by δ = | z (j + 1) −z (j) |. Here, the periodic function obtained using the above equation (4) is also referred to as a P expression. In addition, what is obtained by Fourier transforming the P expression is also referred to as a P-type Fourier descriptor c (k).

From the above equation (4), w (j) can be expressed as a periodic function w (j) having components (j = 1,..., N) on the complex plane as sample points. Further, the component z (j) of the contour C can be expressed using the periodic function w (r) by the following equation (5) developed from the above equation (4).

  That is, the contour line C can be represented by a periodic function w (j), and the two have a one-to-one relationship. This will be described with reference to FIGS. FIG. 10B shows the transition of the change amount of the real part (x) of the contour C developed on the complex plane, and FIG. 10C shows the periodic function w (j) of the real part (x). ). In FIG. 10B, the vertical axis represents the coordinates (x) of the real part (x), and the horizontal axis represents the number of pixels. In FIG. 10C, the vertical axis is the amount of change, and the horizontal axis is the number of pixels. In the present embodiment, since the analysis target is a digital image, the distance between pixels is 1 at the maximum. Therefore, the amount of change (vertical axis) is −1 to 1. As shown in FIGS. 10B and 10C, in the region where the real part (x) has a lot of shakiness, the fluctuation is large even in the periodic function w (j). This indicates that the contour C can be analyzed by frequency analysis on the periodic function w (j).

  Here, when the contour data component is acquired by, for example, a search in the vicinity of 8, it is necessary to perform the following correction. Now, let z (j ') be the component of the contour data searched for in the vicinity of eight. In this case, the distance δ on the complex plane in each component z (j ′) is not always a constant value. This means that the sampling interval is not constant. Therefore, z (j) where the sampling interval is constant while satisfying the value of the deflection angle θ is re-searched based on z (j ′), and the periodic function w (j) is obtained from the obtained z (j). Approximate. Of course, when the contour data component is searched by the 4-neighbor search, the distance δ = 1 on the complex plane is constant, and thus the periodic function w (j) is used by applying the above formula (4) as it is. it can.

In step S33, the Fourier descriptors c (k) (k = 0,..., N-1) of the real part (x) and the imaginary part (yi) of the periodic function w (j) are acquired by the DFTC 433. Specifically, as shown by the following formula (6), the DFTC 433 performs a discrete Fourier transform on the periodic function w (j) to obtain a Fourier descriptor c (k).

  In step S34, the specific frequency component of the Fourier descriptor c (k) (P-type Fourier descriptor) acquired in step S34 is extracted by the band pass filter 434. FIG. 10 (d) shows the Fourier descriptor c (k) in the periodic function w (j). In the diagram shown in FIG. 10D, the horizontal axis is the frequency, and the vertical axis is the frequency component (c (k)).

  In the present embodiment, features of contour data are extracted using a specific frequency component having high sensitivity of human visual characteristics among high frequency components. Here, for example, (−2 to −12 cycle / mm, 2 to 12 cycle / mm) is extracted by the band-pass filter 434 as a frequency component having high sensitivity of human visual characteristics.

  As shown in FIG. 10D, the low frequency component (for example, −2.0 to 2.0 cycle / mm) of the Fourier descriptor c (k) indicates the outline of the contour C, and the high frequency component (for example, , −2.5 cycle / mm or less, 2.5 cycle / mm or more) indicates details. Therefore, the frequency component extracted by the band-pass filter 434 is a frequency component that has high sensitivity to human perception characteristics and is easily affected by the degree of rattling.

  In step S35, the IDFTC 435 performs inverse discrete Fourier transform on the specific frequency component extracted in step S34. By this inverse discrete Fourier transform, a spectrum change corresponding to the specific frequency component in the real part (x) and the imaginary part (yi) of the contour C is acquired.

  FIG. 11 is a waveform diagram showing a spectrum change. As compared from FIG. 11, the spectrum change of the real part (x) and the imaginary part (yi) shown in FIG. 11A is larger than that of FIG. 11B. That is, it can be understood that the spectrum shown in FIG. 11A has a greater degree of rattling than the spectrum shown in FIG.

ここで、ｌは、ステップＳ３５で取得されたスペクトルを識別する値（ｌ＝１ … ｎ’）、ｓ（ｌ）は、スペクトルの値、ａｖ（ｓ）は、スペクトルの平均値、を示す。 Therefore, in step S36, in order to quantitatively show the change of each spectrum described above, the quantification unit 436 uses the standard deviation std (x ′ (t) of each spectrum of the real part (x) and the imaginary part (yi). )), Std (y ′ (t)) is acquired. For example, the standard deviation in the real part (x) is calculated using the following formula (7).

Here, l represents a value (1 = 1... N ′) for identifying the spectrum acquired in step S35, s (l) represents a spectrum value, and av (s) represents an average value of the spectrum.

In step S37, the quantification unit 436 shows the characteristics of the contour data based on the standard deviation (std (x ′ (t)), std (y ′ (t))) acquired in step S36. The rattling amount Rag is calculated. The rattling amount Rag indicates the degree of shading or the degree of unevenness of the contour data. Specifically, the backlash amount Rag is calculated using the following equation (8).

By using the standard deviation of the spectrum as the quantification method, it becomes possible to quantify the amount of rattling more easily.

(4) Index Value Calculation Method Hereinafter, the index value calculation method executed in step S5 of FIG. 3 will be described.
FIG. 12 is a flowchart showing in detail the process of step S5. FIG. 13 is a diagram showing subjective values used in calculating the index value.

  In step S51 of FIG. 12, the rattling amount Rag is acquired by the index value calculation unit 44. In this embodiment, the rattling amount Rag is acquired for each of the three types (20pt, 10.5pt, 4pt) of character scales included in the evaluation image.

  In step S <b> 52, the numerical value of density is acquired by the index value calculation unit 44. The density value is obtained by averaging, for example, the density inside the contour (contour data), the lightness L * inside the contour, or the contrast with paper white by the total number of components (pixels) of the contour data. is there.

In step S53, the index value calculation unit 44 calculates the index value Ind_q. For example, the index value Ind_q is obtained by the following equation (9). In the above example, the index value Ind_q is calculated for each of the three types of character scales.

The coefficients A _{1 to} A ₃ for calculating the index value Ind_q represented by the above formula (9) are calculated using a well-known multivariate analysis. For example, an evaluation image is printed by a plurality of printers, and a subjective value described later is obtained. Then, the correlation is obtained with the subjective value as the objective variable and the backlash amount Rag and the density value OD as the explanatory variables. Based on this correlation, the coefficients A _{1 to} A ₃ in the above equation (9) are acquired. In the present embodiment, the PC 30 records the respective coefficients A _{1 to} A ₃ for each character size. Further, by using the density characteristic in addition to the feature of the contour data, the index value can be calculated according to the individual density characteristic of the evaluation image.

  For example, as shown in FIG. 13, the subjective value is recorded as a value (subjective value Sbj) obtained by a subjective experiment for any 12 types of printers. In FIG. 13, the subjective value Sbj is created for each character size and recorded in the external memory 35. 13A to 13C show subjective values of character sizes of 20 pt, 10.5 pt, and 4 pt. This subjective value Sbj is, for example, a value obtained by an experiment using a one-pair comparison method or the like from an arbitrary subject.

In step S54, the index value calculation unit 44 evaluates the calculated index value Ind_q. For example, the index value calculation unit 44 performs evaluation based on whether or not the index value Ind_q falls within a predetermined standard value.
In addition, the evaluation result may be presented for each character size, or the evaluation result may be a low evaluation when there is at least one character size with a low evaluation. Furthermore, the index value Ind_q for each character size may be averaged, and evaluation may be performed using the averaged index value.

  As described above, according to the present invention, since the evaluation image can be inspected in consideration of human visual characteristics, the processing and apparatus required for the inspection are simplified while maintaining the necessary quality. be able to.

(5) Other Embodiments Configuring the inspection apparatus 1 with the printer 10, the scanner 20, and the PC 30 is merely an example. For example, the function of the inspection apparatus 1 of the present invention may be realized by the printer 10 alone. Furthermore, the inspection apparatus 1 may inspect an image displayed on an LCD (liquid crystal display), or may inspect an image projected on a projector. In this case, for example, an evaluation image is displayed by an LCD or a projector, and an image displayed by an imaging device such as a digital camera is captured. Next, the inspection device 1 acquires a captured image obtained by imaging from the imaging device as an evaluation image. And the inspection apparatus 1 should just perform the process of this invention with respect to this image for evaluation.

  In addition, acquiring the characteristics of the contour data based on the P expression is merely an example. Frequency analysis may be performed based on other expressions (G expressions) or other approximate functions.

  The use of Fourier analysis for frequency analysis is an example. For example, the frequency may be analyzed by wavelet analysis.

  Applying visual characteristics in the binarization process is just one example. For example, the contour data may be extracted only by normal binarization processing. Further, the contour data may be extracted by multivalue processing.

Needless to say, the present invention is not limited to the above embodiments.
That is, the mutually replaceable members and configurations disclosed in the above embodiments may be applied by appropriately changing the combination.
Members and structures that are known techniques and can be mutually replaced with the members and structures disclosed in the above-described embodiments may be appropriately replaced, and combinations thereof may be changed and applied.
Those skilled in the art may appropriately replace the members and structures that can be assumed as substitutes for the members and structures disclosed in the above-described embodiments based on known techniques and the like, and change the combinations thereof.

  DESCRIPTION OF SYMBOLS 1 ... Inspection apparatus, 10 ... Printer, 20 ... Scanner, 30 ... Personal computer (PC), 31 ... Controller, 32, 33 ... IF, 34 ... Keyboard, 35 ... External memory, 36 ... Display, 41 ... Color conversion part, 42 ... contour extraction unit, 43 ... feature acquisition unit, 44 ... index value calculation unit, 421 ... DFTC, 422 ... VTF application unit, 423 ... IDFTC, 424 ... binarization processing unit, 425 ... coordinate extraction unit

Claims (7)

An inspection apparatus for evaluating the quality of an evaluation image,
Image acquisition means for acquiring the evaluation image;
Contour extracting means for extracting contour data indicating a contour included in the acquired evaluation image;
An inspection apparatus comprising: quantification means for performing frequency analysis on the extracted contour data and quantifying characteristics of the contour data based on a spectrum value in a frequency band in which the sensitivity of human visual characteristics is high.   The inspection apparatus according to claim 1, wherein the quantification unit indicates a relative positional relationship of each line segment forming the contour data by a predetermined periodic function, and performs the frequency analysis on the periodic function.   The inspection apparatus according to claim 1, wherein the characteristic of the quantified contour data is acquired based on a deviation of a spectrum value in the frequency component. The contour extracting means multi-values the evaluation image to extract the contour data,
The inspection apparatus according to any one of claims 1 to 3, wherein the threshold value used in the multi-value conversion is set in consideration of human visual characteristics.   The inspection according to any one of claims 1 to 4, further comprising index value calculation means for calculating an index value indicating the quality of the evaluation image based on the quantified characteristic of the contour data. apparatus.   6. The inspection apparatus according to claim 5, wherein the index value calculation means calculates the index value based on the quantified characteristic of the contour data and a density characteristic of a portion surrounded by the contour. An image acquisition step of acquiring an evaluation image;
A contour extraction step of extracting contour data indicating a contour included in the acquired evaluation image;
And a quantification step of performing a frequency analysis on the extracted contour data and quantifying the characteristics of the contour data based on a spectrum value in a frequency band in which the sensitivity of human visual characteristics is high.

Images