JPH09204529A - Image processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a nice-looking contour and also to effectively carry out the desired image processing in a short time by deciding the type of an input image and extracting the contour of the input image by means of the contour extraction method that is accordant with the type of the input image. SOLUTION: An original image is picked up by a digital still camera 60 and this picked up image is sent to a personal computer 69. The computer 69 displays the received original image on a display, etc., serving as an output device 67. An operator decides whether the displayed image is identical with a machine drawing, a landscape image or a figure image and then inputs this decision result via a mouse keyboard, etc., serving as an input device 68. An image type decision part 4 of a contour extraction mode selection part 3 sends the type signal of the image to be processed to a contour extraction part 8 together with the original image data. The part 8 decides the contour extraction method based on the inputted image type.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus, and more particularly, to extracting a contour line from a grayscale image input by a digital still camera, a scanner or the like to create a line figure used for an instruction manual or a manual. The present invention relates to an image processing device.

[0002]

2. Description of the Related Art As a technique for extracting a contour line from a grayscale image input by a digital still camera, a scanner or the like to create a line figure to be used in an instruction manual, a manual, etc., Japanese Patent Laid-Open No. 61-175879. Japanese Patent Laid-Open No. 1-295378 and Japanese Patent Laid-Open No. 1-295378 are disclosed. Both of them employ a method of differentiating the input data in the original image in a predetermined direction to extract the contour line of the original image.

[0003]

However, in any of the above-mentioned conventional techniques, it is not possible to simply consider what kind of image the input original image is and what kind of property it has. Since the contour line is extracted using one contour line extraction method, the appearance of the extracted contour line image is not good. In addition, when the input original image is a large image or the number of original images is very large, the processing time becomes unnecessarily long and time-consuming, which is not efficient.

The image processing apparatus of the present invention has been made in view of such a problem, and an object thereof is to obtain a contour line image having a good appearance and to efficiently perform desired image processing. An object of the present invention is to provide an image processing device that can be performed in a short time.

[0005]

In order to achieve the above object, the image processing apparatus of the present invention displays an image input section for inputting an image and an input image input from the image input section. For determining the type of the input image based on the information obtained from the input image displayed on the image display unit, and the contour line according to the determination result by the type determining unit. And a contour line extracting means for extracting the contour line of the input image by using the extraction method.

Further, the image processing apparatus of the present invention further comprises a processing area designating section for designating an area where contour line extraction is to be performed on the input image displayed on the image display section.

Further, the type determining means includes a Fourier transform unit for converting an input image into Fourier coefficient data,
A spectrum calculation unit for converting the obtained Fourier coefficient data into spectrum data, a correlation calculation unit for performing a correlation calculation between the obtained spectrum data and the template spectrum data for classifying the input image, and And a correlation coefficient determining section for determining the correlation coefficient calculated by the correlation calculating section.

That is, the image processing apparatus of the present invention displays the input image input from the image input section on the image display section,
The type determining means determines the type of the input image based on the information obtained from the displayed input image. And
The contour line extracting method extracts the contour line of the input image by using the contour line extracting method according to the determination result by the type determining means.

Further, in the image processing apparatus of the present invention, the processing area designating section displayed on the image display section designates an area to be subjected to contour line extraction for the input image.

In the type determining means, the input image is converted into Fourier coefficient data in the Fourier transform section,
The obtained Fourier coefficient data is converted into spectrum data in the spectrum calculation unit, and the correlation calculation unit performs the correlation calculation between the obtained spectrum data and the template spectrum data for classifying the input image. The correlation coefficient is determined by the correlation coefficient determination unit.

[0011]

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

First, the outline of this embodiment will be described with reference to FIG. The image input from the image input unit 1 is displayed on the original image display unit 2. The operator determines what kind of image the displayed image is, and inputs the determination result to the contour line extraction mode selection unit 3. The contour line extraction mode selection unit 3 determines the type of the input image based on the input determination result and selects the optimum contour line extraction mode. The contour line extraction unit 8 extracts the contour line of the input image by a contour line extraction method suitable for the selected contour line extraction mode and creates a contour line image. The created contour image is displayed on the processed image display unit 9.

The first embodiment of the present invention will be described below.
FIG. 2 is a diagram showing the configuration of the first embodiment. The digital still camera 60 corresponds to the image input unit 1 in FIG. 1 and can be configured by a scanner or the like. Output device 6
Reference numeral 7 corresponds to the original image display unit 2 and the processed image display unit 9 in FIG. 1, and is constituted by a display or the like. The original image display unit 2 displays the original image input by the image input unit 1,
The processed image display unit 9 displays the processed image that has been processed or contour line extracted. The memory 63 stores the processing procedures of the contour line extraction mode selection unit 3 and the contour line extraction unit 8. A contour line extraction mode selection unit 3, a contour line extraction unit 8,
The memory 63 and the frame memory 64 form a personal computer 69.

Further, in the first embodiment, the contour line extraction mode selection unit 3 shown in FIG. 1 has an image type determination unit 4 for determining the image type, as shown in FIG. The type of the original image is determined by outputting the type signals a (5), b (6), and c (7) expressing the image type as a result of the determination by the image type determination unit 4. Also,
As shown in FIG. 4, the contour line extraction unit 8 shown in FIG. 1 has a type signal determination unit 10 and determines the type signal. Based on this determination result, (1) the differential processing unit 11 that uses the differential value of the brightness change of the image, and (2) the template that assumes the grayscale pattern near the outline on the image Template Method Execution Unit 12 for Detecting Contour Elements, (3) Can which performs noise removal and thinning by blurring the contour points with a Gaussian function and then calculating the gradient
Any one of the ny method execution units 13 is selected and executed. The differential processing (1) is disclosed in, for example, the Image Analysis Handbook, The University of Tokyo Press, 1992, pp 550-554.
The template method of (2) is described in, for example, Image Analysis Handbook, The University of Tokyo Press, 1992, pp.
554-555. Moreover, Can of (3)
For ny's method, see IEEE TRANSACTI
ONS ON PATTERN ANALYSIS A
ND MACHINE INTELLIGENCE, V
OL PAMI-8 NO. 6,1986, pp679
-698.

The operation of the above configuration will be described below. First, in FIG. 2, an original image is captured by the digital still camera 60. The captured original image is the storage medium 61.
Is stored. This original image is read by the data reading device 62 connected to the personal computer 69 and transferred to the personal computer 69. The transferred original image is read into the frame memory 64 and displayed on the display or the like which is the output device 67. The operator determines whether the image displayed on the display is a mechanical drawing, a landscape image, or a person image, and inputs the determination result using the mouse, keyboard, or the like which is the input device 68.

In the image type determination unit 4 in the contour line extraction mode selection unit 3, in response to an input from the operator, the type signals of the images to be processed are sent to the contour line extraction unit 8 as the type signals a to c together with the original image data. send. In the contour line extraction unit 8, first,
The type signal determination unit 10 determines the type signal. When the type signal is a, the method 1 is selected, the original image data is sent to the differentiation processing section 11, and the contour line is extracted.
When the type signal is b, the method 2 is selected, the original image data is sent to the template method execution unit 12, and the contour line is extracted. When the type signal is c, the method 3 is selected, the original image data is sent to the Canny method execution unit 13, and the contour line is extracted. The contour line extraction method is determined according to the type of the image thus input. The image data from which the contour has been extracted is sent again to the output device 67 and displayed on the display. Further, the image data from which the contour line has been extracted is transferred to the external storage device 66, and the processing result is stored.

In the above-described first embodiment, the original image input by the digital still camera 60 is displayed on the output device 67, the operator looks at the displayed image to confirm the type of the input image, and then the input image is input. The type of image is input using the device 68. Then, the contour line extraction unit 8 extracts the contour line of the input image by using the contour line extraction method suitable for the original image based on the type signal, so that the appearance of the extracted contour line image is improved.

The second embodiment of the present invention will be described below.
As shown in FIG. 5, the second embodiment is different from the first embodiment in that a processing area designation unit 14 is provided between the original image display unit 2 and the contour line extraction mode selection unit 3 shown in FIG.

The processing area designating section 14 is for designating a processing area in the original image by a mouse, a keyboard or the like which is the input device 68 shown in FIG.

The operation of the second embodiment will be described below. In FIG. 2, an original image is captured by the digital still camera 60. The captured original image is stored in the storage medium 61. This original image is read by the data reading device 62 connected to the personal computer 69 and is read by the personal computer 6.
9 is transferred. The transferred original image is read into the frame memory 64 and displayed on the display or the like which is the output device 67.

Next, as shown in FIG. 6, the operator uses the input device 68 to select the processing area 15 for the original image 16 displayed on the display. The operator subsequently determines whether the original image is a mechanical drawing, a person image, or a landscape image, and the input device 6
The judgment result is input by 8. The image type determination unit 4 in the contour line extraction mode selection unit 3 sends the type signals of the images to be processed as the type signals a to c together with the data of the processing area to the contour line extraction unit 8 in response to the operator's input. In the contour line extraction unit 8, first, the type signal determination unit 10 determines the type signal. When the type signal is a, the method 1 is selected, the original image data is sent to the differentiation processing section 11, and the contour line is extracted. When the type signal is b, the method 2 is selected, the original image data is sent to the template method execution unit 12, and the contour line is extracted. Further, when the type signal is c, the method 3 is selected, the original image data is sent to the Canny's method execution unit 13, and the contour line is extracted. The processing area data from which the contour line has been extracted is sent again to the output device 67 and displayed on the display. Further, the processing area data from which the contour line has been extracted is transferred to the external storage device 66 and the processing result is stored.

According to the above-described second embodiment, since the predetermined area is selected from the original image, the operator can obtain only the necessary information. Further, when the original image is very large, the area to be processed can be limited by selecting a predetermined area, which can shorten the processing time and is efficient.

The third embodiment of the present invention will be described below.
In the third embodiment, as shown in FIG. 7, a preprocessing unit 17 is provided between the original image display unit 2 and the contour line extraction mode selection unit 3 shown in FIG.
The point that is provided is different from the first embodiment.

The preprocessing section 17 removes noise contained in the input image, for example. As the noise removal method performed by the preprocessing unit 17, at least one of a smoothing filter, a Gaussian filter, a method of cutting high frequency components by using a Fourier transform, and contour line preserving smoothing is used. The smoothing filter obtains the average value of the peripheral area centered on the pixel of interest and sets it as the value of the central pixel. The Gaussian filter weights the image with a two-dimensional Gaussian function. The method of cutting the high frequency component by using the Fourier transform applies a Fourier transform to the image and applies a filter that cuts off the high frequency component in the Fourier plane, and inversely transforms it.
The contour-preserving smoothing is performed while sharpening the contour information in the image.

The operation of the third embodiment will be described below. In FIG. 2, an original image is captured by the digital still camera 60. The captured original image is stored in the storage medium 61. This original image is read by the data reading device 62 connected to the personal computer 69 and transferred to the personal computer 69. The transferred original image is read into the frame memory 64 and displayed on the display or the like which is the output device 67. The operator uses the input device 68 to select the noise removal method for the original image 16 displayed on the display. Then, it is determined whether the original image is a mechanical drawing, a person image, or a landscape image. The operator inputs the judgment result by the input device 68.

In the image type determination unit 4 in the contour line extraction mode selection unit 3, the type signals of the image to be processed are sent to the contour line extraction unit 8 together with the data of the processing area as the type signals a to c in response to the operator's input. send. In the contour line extraction unit 8, first,
The type signal determination unit 10 determines the type signal. When the type signal is a, the method 1 is selected, the original image data is sent to the differentiation processing section 11, and the contour line is extracted.
When the type signal is b, the method 2 is selected, the original image data is sent to the template method execution unit 12, and the contour line is extracted. When the type signal is c, the method 3 is selected,
The original image data is sent to the Canny technique execution unit 13 to perform contour line extraction. The processing area data from which the contour line has been extracted is sent again to the output device 67 and displayed on the display. Further, the processing data from which the contour line has been extracted is transferred to the external storage device 66, and the processing result is stored.

According to the above-described third embodiment, the noise contained in the input image is removed by the preprocessing unit 17, so that the output image can be generated as a sharp image with less noise, and the contour line can be generated. After the extraction, it is possible to greatly reduce the labor of the manual correction of the operator.

The fourth embodiment of the present invention will be described below.
As shown in FIG. 8, the embodiment of FIG. 4 differs from the first embodiment in that a post-processing unit 40 is provided between the contour line extraction unit 8 and the processed image display unit 9 shown in FIG. FIG. 9 is a view showing the arrangement of the post-processing unit 40. At least one of noise removal, thinning, and binarization is executed on the image whose contour line has been extracted by the contour line extraction unit 8. It is configured. Reference numeral 18 denotes a selection unit for selecting whether or not to remove noise, and 19
Represents a noise removal processing unit. For the noise removal processing, a median filter for obtaining the median value of pixel values in the neighborhood area, or erosion or dilation which is a method of expanding or contracting the original figure is used.

Reference numeral 20 denotes a selection unit for selecting whether or not to perform thinning processing, and 21 denotes a thinning processing unit for thinning the contour line information in the image. In the thinning process, in the case of a multi-valued image, the differential value at the pixel of interest is compared with the differential amount of a neighboring element adjacent in the direction perpendicular to the contour line, and the contour having a smaller differential value than the neighboring Remove lines nonmaximasup
The compression method is used, and in the case of a binary image, the binary image thinning method is used. Reference numeral 22 represents a selection unit that selects whether or not to perform binarization, and 23 represents a binarization processing unit that binarizes the image. For the binarization processing, a binarization method is used in which an optimum threshold value is obtained and the density histogram of the image is divided into two classes.

The operation of the fourth embodiment will be described below. In FIG. 2, an original image is captured by the digital still camera 60. The captured original image is stored in the storage medium 61. This original image is read by the data reading device 62 connected to the personal computer and is read by the personal computer 69.
Is forwarded to. The transferred original image is the frame memory 6
4 and is displayed on the output device 67 such as a display. The operator determines whether the original image 16 displayed on the display is a mechanical drawing, a person image, or a landscape image, and makes an input using the input device 68.

The image type determining unit 4 sends the type signals of the image to be processed by the operator's input to the contour line extracting unit 8 as the type signals a to c together with the data of the processing area. In the contour line extraction unit 8, first, the type signal determination unit 10 determines the type signal. When the type signal is a, the method 1 is selected, the original image data is sent to the differentiation processing section 11, and the contour line is extracted. When the type signal is b, the method 2 is selected, the original image data is sent to the template method execution unit 12, and the contour line is extracted. Further, when the type signal is c, the method 3 is selected, the original image data is sent to the Canny's method execution unit 13, and the contour line is extracted. When the noise removal processing 19 is necessary for the image from which the contour line has been extracted, the operator inputs it using the input device 68, and the noise removal processing is performed.

FIGS. 10A, 10B, and 10C are diagrams showing the states of contour line extraction and noise removal processing. FIG.
Three regions A, B, C having different densities as shown in (a)
FIG. 10B shows an image obtained by performing the contour line extraction process on the original image consisting of. Since the image in FIG. 10B generally includes noise (a portion indicated by an x mark), it is necessary to perform noise removal processing in post processing. FIG. 10C shows the result of the noise removal processing performed on FIG.

Further, when it is necessary for the thinning processing unit 21 to perform thinning on the image from which the contour has been extracted, the operator uses the input device 68 to input, and thinning processing is performed. 11A and 11B show the thinning process. Although the line width of the extracted contour line is different in FIG. 11A before performing the thinning, the line width can be unified as shown in FIG. 11B by the thinning process. In addition, it is 2
When the binarization processing in the binarization processing unit 23 is necessary,
The operator uses the input device 68 to input, and binarization processing is performed. The processed data subjected to the post-processing is sent again to the output data 67 and displayed on the display. Further, the processed data is transferred to the external storage device 66, and the processing result is stored.

According to the above-described fourth embodiment, the post-processing unit 17 selectively performs noise removal, thinning, binarization, etc., so that the output image is a sharp image with little noise. Will be able to generate. In addition, the thinning of the line improves the expressiveness of the illustration image. Further, by performing binarization, it becomes possible to reduce the amount of data when saving an image.

The fifth embodiment of the present invention will be described below.
In the fifth embodiment, the configuration of the post-processing unit of the fourth embodiment is replaced with another configuration. That is, in FIG. 12, at least one of noise removal and thinning is performed on the image from which the contour line is extracted by the contour line extraction unit 6, and then the binarization processing unit 23 performs binarization processing, Further, the line segment component correction unit 50 is configured to correct the line segment component.

FIG. 13 shows the construction of the line segment component correction unit 50, and 25 is a Hough transform unit for performing Hough transform on the input image. In the Hough transform, as shown in FIG. 14, the length of the perpendicular line ρ drawn from the origin O to the line segment k and the angle θ formed by the perpendicular line ρ and the x-axis, the coordinates on the screen are expressed with ρ and θ as parameters. You can One point on the image forms one curve on the ρ-θ plane (Huff plane), and one point on the Huff plane represents one line segment on the image. In other words, for both
= Xcos θ + y sin θ holds.

On the Huff plane, a point where many curves intersect represents a line segment. Therefore, it is necessary to set a threshold value in order to extract the line segment. Therefore, as shown in FIG. 27 is provided. Reference numeral 28 is an inverse conversion unit that converts the value extracted by the threshold setting unit 27 into image data. Reference numeral 29 is a constraint condition setting unit for reproducing an accurate line segment from the inversely transformed image.

The constraint condition setting unit 2 will be described below with reference to FIG.
The process of 9 will be described. 15A, 15B, and 15C are diagrams showing an example of the Hough transform operation. FIG. 15A is a contour line extraction image, which includes various components such as line segments and curves. In order to extract only the line segment components, a certain threshold value is set, Hough transform is performed, and the result of the inverse transform is shown in FIG. By attaching a constraint condition to extract only the line segment corresponding to the line segment of the original image, the line segment shown in FIG.
As shown in (c), only line segment information can be extracted.

The line segment A'B 'in FIG. 15B is shown in FIG.
This corresponds to the line segment AB of the original image in (a). Here, points A and B in the original image are end points. A method is used in which the endpoint information is detected in the inversely transformed image based on the line segment in the corresponding original image and the information is used as a constraint condition.

For example, the center coordinates C (x0, y0) of the line segment AB are calculated from the end points A and B on the line segment AB in FIG. 15 (d). The coordinates of the end points A and B are A (x1, y1),
If B (x2, y2), the center coordinate C of the AB line is x0
= (X1 + x2) / 2, y0 = (y1 + y2) / 2 can be calculated. The distance between points A and C is D =
^{{(X0 -x1) 2 + (} y0 -y1) 2} can be obtained from the ^half of the equation.

Here, the point corresponding to the center coordinate C on the line segment AB of FIG. 15D is defined as C ′ on the line segment A′B ′ of FIG. 15B. Let any point on the line segment A'B 'be E' (x,
y), the distance of the line segment C′E ′ is D ′ = {(xx
^{0) 2 + (y-y0} ) 2} can be calculated ^1/2. If D ′ is larger than D, E ′ does not exist in the original image and the point E is not drawn. When D ′ is D or less, the point E is drawn as if the E ′ exists in the original image. By applying such a constraint condition, the end point A
And only the line segment within the range of the end point B is extracted. An image as shown in FIG. 15C is obtained by applying such a constraint condition to all the line segments extracted by the Hough transform.

According to the fifth embodiment described above, by providing the post-processing unit 40 with the line-segment component correction unit 50 using Hough transform, it is possible to correct unclear line-segment components at the time of image pickup, and You will be able to extract only the minutes.

The sixth embodiment of the present invention will be described below.
The sixth embodiment uses the configuration shown in FIG. 16 as the configuration of the contour line extraction mode selection unit 3 shown in FIG.

In FIG. 16, the original image data is converted into Fourier coefficient data by the Fourier transform unit 29.
The spectrum calculation unit 30 converts the Fourier coefficient data into spectrum data. The correlation calculator 32 calculates a correlation coefficient between the spectrum data and the template spectrum data from the template spectrum data generator 31. Here, three template spectrum data are prepared: mechanical drawing template data, portrait image template data, and landscape image template data. As the template spectrum data, the result of statistically processing the spectrum data calculated from a plurality of mechanical drawing images, portrait images, and landscape images prepared in advance is used. The correlation coefficient is calculated for each image in the correlation calculation unit 32, and the three obtained correlation coefficients are sent to the correlation coefficient determination unit 33.

The correlation coefficient determination unit 33 determines the correlation coefficient using the table shown in FIG. In FIG. 17, a correlation coefficient A is the mechanical drawing template data and the original image, a correlation coefficient B is the human image template data and the original image, and a correlation coefficient C is the landscape image template data and the original image. It represents. In (1), when the correlation coefficient A is the maximum and is 0.5 or more, the determination result is a and the type signal a is sent to the contour line extraction unit 8. In (2), when the correlation coefficient B is the maximum and is 0.5 or more, the determination result is c, and the type signal b is sent to the contour line extraction unit 8. In (3), when the correlation coefficient C is maximum and is 0.5 or more, the determination result is c, and the type signal c is sent to the contour line extraction unit 8. In (4), when none of the correlation coefficients is less than 0.5, the type signal d is sent to the contour line extraction unit 8.

Next, in the contour line extracting section 8 (FIG. 4), first,
The type signal determination unit 10 determines the type signal. When the type signals are a and d, the method 1 is selected, the original image data is sent to the differentiation processing unit 11, and the contour line is extracted. When the type signal is b, the method 2 is selected, the original image data is sent to the template method execution unit 12, and the contour line is extracted. Further, when the type signal is c, the method 3 is selected, the original image data is sent to the Canny's method execution unit 13, and the contour line is extracted.

In this embodiment, the template spectrum data 31 is set to three, but the image types may be classified more finely to prepare three or more template spectrum data 31.

In the sixth embodiment described above, the Fourier transform unit 29 and the spectrum calculation unit 30 calculate the spectrum data of the original image, and the correlation coefficient with the template spectrum data is obtained to automatically determine the type of the image. Judge, select the contour extraction method suitable for that image,
The contour line is extracted. Therefore, for example, when the number of original images is very large, the operator can automatically extract the contour line of the image only by designating the image to be processed, which improves the work efficiency.

The seventh embodiment of the present invention will be described below. The seventh embodiment uses a configuration shown in FIG. 19 instead of the contour line extraction mode selection unit 3 described in the sixth embodiment. In FIG. 19, the original image data is converted into Fourier coefficient data by the Fourier transform unit 29. The spectrum calculation unit 30 converts the Fourier coefficient data into spectrum data. The spectrum data is blocked by the block unit 37.
Is input to and block processing is performed.

FIG. 18 is an explanatory diagram of the blocking process.
First, with respect to the spectrum data 38, the average value of four spectra having close frequency bands is obtained. Here, four averages are obtained, but the number is not limited to four, and any number may be used. The correlation calculator 32 calculates a correlation coefficient between the spectrum data and the template spectrum data from the template spectrum data generator 31. For example, if three template spectrum data, machine drawing template data, portrait image template data, and landscape image template data, are prepared, the correlation coefficient is calculated for each image, and the three obtained correlation coefficients are obtained. Is sent to the correlation coefficient determination unit 33.

The correlation coefficient determination unit 33 determines the correlation coefficient using the table shown in FIG. The correlation coefficient A represents the mechanical drawing template data and the original image, the correlation coefficient B represents the portrait image template data and the original image, and the correlation coefficient C represents the calculation result of the landscape image template data and the original image. In (1), when the correlation coefficient A is the maximum and is 0.5 or more, the determination result is a and the type signal a is sent to the contour line extraction unit 8. In (2), when the correlation coefficient B is the maximum and is 0.5 or more, the determination result is c and the type signal b is the contour line extraction unit 8
Sent to In (3), when the correlation coefficient C is maximum and is 0.5 or more, the determination result is c, and the type signal c is sent to the contour line extraction unit 8. In (4), when none of the correlation coefficients is less than 0.5, the type signal d is sent to the contour line extraction unit 8.

Next, in the contour line extraction unit 8 (FIG. 4), first,
The type signal determination unit 10 determines the type signal. When the type signals are a and d, the method 1 is selected, the original image data is sent to the differentiating process 11, and the contour line is extracted. When the type signal is b, the method 2 is selected, the original image data is sent to the template method execution unit 12, and the contour line is extracted. Further, when the type signal is c, the method 3 is selected, the original image data is sent to the Canny's method execution unit 13, and the contour line is extracted.

In the seventh embodiment described above, since the original image is divided into blocks, the number of data can be reduced. Therefore, the correlation calculation can be performed at high speed, and the processing time can be shortened.

The specific embodiments described above include inventions having the following configurations, and the effects of each configuration are as follows.

Structure (1) An image input section for inputting an image, an image display section for displaying an input image input from the image input section,
Based on the information obtained from the input image displayed on the image display unit, the input image is obtained by using a type determining means for determining the type of the input image and a contour line extracting method according to the determination result by the type determining means. An image processing apparatus, comprising: a contour line extracting unit that extracts the contour line of the image processing apparatus.

(Effect) The input original image is displayed on the image display section, and the operator looks at the displayed image to confirm the type of the input image, and then inputs the type of the image. Since the contour line extraction unit is configured to extract the contour line of the input image using the contour line extraction method suitable for the original image based on the type signal, the appearance of the extracted contour line image is improved.

Structure (2) The image processing according to structure (1), further comprising a processing region designating unit for designating a region in which contour line extraction is to be performed on the input image displayed on the image display unit. apparatus.

(Effect) Since the predetermined area is selected from the original image, the operator can obtain only necessary information. Further, when the original image is very large, the area to be processed can be limited by selecting a predetermined area, which can shorten the processing time and is efficient.

Structure (3) The image processing according to structure (1), further comprising a pre-processing unit for removing a noise component in the input image before the type judging unit judges the type of the input image. apparatus.

(Effect) Since the noise contained in the input image is removed by the preprocessing unit, the output image can be generated as a sharp image with less noise, and the correction is manually performed by the operator after the contour line extraction. The labor of can be reduced significantly.

Structure (4) After the contour line of the input image is extracted by the contour line extraction means, at least one process of noise removal, thinning, and binarization is performed on the input image from which the contour line is extracted. The image processing apparatus according to (1), further including a post-processing unit that performs the processing.

(Effect) Since noise removal, thinning, binarization and the like are selectively performed in the post-processing section, an output image can be generated as a sharp image with little noise. In addition, the expression of the illustration image is improved by performing the thinning. Further, by performing binarization, it becomes possible to reduce the amount of data when saving an image.

Structure (5) The image processing according to structure (4), wherein the post-processing unit includes a line segment component correction unit for extracting a line segment component and correcting the line segment component. apparatus.

(Effect) By providing a line segment component correction unit using Hough transform in the post-processing unit, it is possible to correct a line segment component which is not clear at the time of image pickup and to extract only the line segment. become.

Configuration (6) The type determining means is obtained by a Fourier transform unit for converting an input image into Fourier coefficient data, a spectrum operation unit for converting the obtained Fourier coefficient data into spectrum data, and And a correlation calculation unit for performing a correlation calculation between the template spectrum data for classifying the input image, and a correlation coefficient determination unit for determining the correlation coefficient calculated by the correlation calculation unit. The image processing device according to configuration (1), including:

(Effect) The spectrum data of the original image is calculated by the Fourier transform unit and the spectrum calculation unit, and the type of the image is automatically determined by obtaining the correlation coefficient with the template spectrum data, which is suitable for the image. The contour line extraction method is selected and the contour line is extracted. Therefore, for example, when the number of original images is very large, the operator can automatically extract the contour line of the image only by designating the image to be processed, which improves the work efficiency.

Structure (7) The image processing apparatus according to structure (6), wherein the type determining means includes a blocking unit for blocking the spectrum data.

(Effect) Since the original image is divided into blocks, the number of data can be reduced. Therefore, the correlation calculation can be performed at high speed, and the processing time can be shortened.

[0069]

As described above, according to the present invention, a contour line image having a good appearance can be obtained, and desired image processing can be efficiently performed in a short time.

[Brief description of drawings]

FIG. 1 is a diagram for explaining the outline of the present embodiment.

FIG. 2 is a diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 3 is a diagram showing a configuration of a contour line extraction mode selection unit shown in FIG.

FIG. 4 is a diagram showing a configuration of a contour line extraction unit shown in FIG.

FIG. 5 is a diagram showing a configuration of a second embodiment of the present invention.

FIG. 6 is a diagram showing how a processing area is designated.

FIG. 7 is a diagram showing a configuration of a third exemplary embodiment of the present invention.

FIG. 8 is a diagram showing a configuration of a fourth embodiment of the present invention.

9 is a diagram showing a configuration of a post-processing unit 40 shown in FIG.

FIG. 10 is a diagram illustrating a state of contour line extraction and noise removal processing.

FIG. 11 is a diagram showing how thinning processing is performed.

FIG. 12 is a diagram showing a configuration of a post-processing unit in a fifth embodiment of the present invention.

FIG. 13 is a diagram showing a configuration of a line segment component correction unit in a fifth embodiment of the present invention.

FIG. 14 is a diagram for explaining Hough transform.

FIG. 15 is a diagram showing an example of a Hough transform operation.

FIG. 16 is a diagram showing a configuration of a sixth exemplary embodiment of the present invention.

FIG. 17 is a table used for correlation coefficient determination.

FIG. 18 is an explanatory diagram of blocking processing.

FIG. 19 is a diagram showing a configuration of a seventh exemplary embodiment of the present invention.

[Explanation of symbols]

1 ... Image input unit, 2 ... Original image display unit, 3 ... Contour line extraction mode selection unit, 8 ... Contour line extraction unit, 9 ... Processed image display unit.

Claims (3)

[Claims] 1. An image input section for inputting an image, an image display section for displaying an input image input from the image input section, and an image obtained from the input image displayed on the image display section. A type determining means for determining the type of the input image based on the information; and a contour line extracting means for extracting the contour line of the input image by using the contour line extracting method according to the determination result by the type determining means. An image processing device characterized by: 2. The image according to claim 1, further comprising a processing region designating unit for designating a region in which contour line extraction is to be performed on the input image displayed on the image display unit. Processing equipment. 3. The type determining means includes a Fourier transform unit for converting an input image into Fourier coefficient data, a spectrum calculation unit for converting the obtained Fourier coefficient data into spectrum data, and the obtained spectrum. A correlation calculation unit that performs a correlation calculation between the data and the template spectrum data for classifying the input image; and a correlation coefficient determination unit that determines the correlation coefficient calculated by this correlation calculation unit. The image processing apparatus according to claim 1, further comprising: