JP2009259036A - Image processing device, image processing method, image processing program, recording medium, and image processing system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing device, an image processing method, an image processing program, a recording medium and an image processing system by which the center of a circle can be decided highly precisely in a short period of time when obtaining the circle center of image data. <P>SOLUTION: A Hough conversion is applied to the image data, and the central position of the detected circle is temporarily set (S2). A prescribed range with the coordinate of the temporarily set circle center position as a center is set, and the circles detected with respective coordinates existing in the prescribed range as original points are converted by a polar coordinate (S5). The radii r of the circles which are obtained by converting by polar-coordinate the circles detected with the respective coordinates as the original points are one-dimensionally projected in the axial direction of a center angle θ (S6). Evaluation values indicating the linearity of the radii r of the circles in the axial direction of the center angle θ are acquired from distribution in the axial direction of the center angle θ concerning the radii r of the circles obtained by the one-dimensional projection (S7). The coordinate of the original point with the highest one of the respective evaluation values from the distribution of the radii r of the circles which are obtained by converting by polar coordinate the circles detected with the respective coordinates existing in the prescribed range as the original points is determined as the circle center position (S9). <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing device, an image processing method, an image processing program, and an image processing system that perform graphic processing on image data input from an image input device such as an imaging device. The present invention relates to an image processing apparatus, an image processing method, an image processing program, a recording medium, and an image processing system that determine a center position of a circular image included.

  An image that images products and parts with a camera and performs various image processing on the obtained image data in order to realize automation of product assembly, automation of inspection work, position adjustment of product assembly, etc. with high accuracy A processing device is required.

  In particular, when the product casing or part has a circular part and this part is used as an alignment mark or inspection part during assembly, the center of the circle is determined by image processing, and the determined center position is determined. As a reference, parts are moved, adjusted, and inspected.

  However, there are chips and burrs in the housing and parts. Therefore, in the captured image data, for example, as shown in FIG. 24A, a part of the circle is missing, or as shown in FIG. 24B, foreign matter, scratches, noise, etc. on the circumference F101. May exist or the circle may be interrupted as shown in FIG. In addition, due to the relative positional relationship between the illumination and the casing or components, as shown in FIG. 24D, unevenness in brightness may occur, and image data with a part of a circle missing may be obtained. . In these cases, the accuracy of determining the center position of the circle is greatly reduced. Therefore, various techniques for determining the center position of the circle of the image data with high accuracy have been developed.

  For example, in Patent Document 1, differentiation processing is performed on a target pixel of image data to calculate an edge strength and an edge vector. This edge vector is calculated using the horizontal edge strength and the vertical edge strength for 3 × 3 pixels centered on the pixel of interest.

  Then, it is assumed that a pixel of interest whose edge intensity is greater than a predetermined threshold is a pixel on the circumference, and for the pixel of interest of the pixel on the circumference, a temporary center of the circle is calculated from the edge vector. Next, the number of pixels present on the circumference passing through the temporary center with the target pixel as the center is counted. These processes are performed for all pixels of interest that are larger than a predetermined threshold, and a maximum circle is provisionally determined. By determining the coordinates of the center of gravity including the pixels within the predetermined width of the tentatively determined circle as the circle center coordinates, the accuracy of determining the circle is improved.

Further, in Patent Document 2, an edge point is detected by differential filter processing for image data captured by a camera, and an arc approximation is performed from the obtained set of edge points by a mean square error minimum method to estimate a circle center position. The set of edge points excluding the points away from the estimated circumference is again subjected to arc approximation by the mean square error minimum method to estimate the circle center position.
Japanese Patent Laying-Open No. 2005-322043 (published on November 17, 2005) Japanese Patent Laid-Open No. 7-225843 (published on August 22, 1995)

  However, the conventional image processing apparatus and image processing method have the following problems.

  First, using the technique disclosed in Patent Document 1, differentiation processing and edge direction calculation are performed for all target pixels, and all pixels that are equal to or greater than a predetermined threshold are on the same circumference. Since it is necessary to determine whether or not there is a problem, the calculation processing time becomes very long. In addition, since the differential processing is easily affected by noise or the like, there is a problem that the calculated value of the edge direction after processing is not accurate.

  Moreover, in patent document 2, since the circular approximation is performed twice with respect to the differential process data which is easy to receive the influence of an error, there exists a problem that the influence of noise etc. is large and detection accuracy is not sufficient.

  Furthermore, since both Patent Document 1 and Patent Document 2 perform differentiation processing in a simple orthogonal coordinate system, a case where a part of a circle is missing due to the influence of a foreign object or the like, or uneven illumination (shading) ) Appears, the detection accuracy cannot be ensured.

  The present invention has been made in view of the above-described conventional problems, and an object of the present invention is to provide an image processing apparatus, an image, and an image processing apparatus capable of accurately determining the center of a circle in a short time when determining the center of the circle of image data. A processing method, an image processing program, a recording medium, and an image processing system are provided.

  In order to solve the above problems, an image processing apparatus according to the present invention is an image processing apparatus that obtains the center position of a circle displayed by image data, which is detected by performing a Hough transform on the image data. A circle center temporary setting means for temporarily setting the center position of the circle and a predetermined range centered on the coordinates of the center position of the temporarily set circle are set, and each of the coordinates existing in the predetermined range is detected as the origin. Polar coordinate conversion means for converting the circle obtained into polar coordinates, and one-dimensional projection means for one-dimensionally projecting the radius r of the circle obtained when the detected circle is converted into polar coordinates using the coordinates as the origin in the direction of the central angle θ axis, respectively. An evaluation value for obtaining an evaluation value indicating linearity of the radius r of the circle with respect to the direction of the central angle θ-axis from the distribution in the direction of the central angle θ-axis at the radius r of the circle obtained by the one-dimensional projection by the one-dimensional projection means. Calculation means and above Among the evaluation values from the distribution of the radius r of the circle obtained by performing polar coordinate conversion on the detected circle with the coordinates existing in a fixed range as the origin, the coordinate of the origin where the evaluation value is highest is the center of the circle. A circle center determining means for determining the position is provided.

  An image processing method according to the present invention is an image processing method for obtaining a center position of a circle displayed by image data in order to solve the above-described problem, and the image data is detected by performing a Hough transform on the image data. The circle center temporary setting step for temporarily setting the center position of the circle, and a predetermined range centered on the coordinates of the center position of the temporarily set circle are set, and each of the coordinates existing in the predetermined range is detected as the origin. A polar coordinate conversion step of converting the circle that has been converted into polar coordinates, and a one-dimensional projection step of performing one-dimensional projection of the radius r of the circle obtained when the detected circle is converted into polar coordinates using the coordinates as the origin in the direction of the central angle θ. An evaluation value for obtaining an evaluation value indicating linearity with respect to the central angle θ-axis direction of the radius r of the circle from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step. Calculation process and above Among the evaluation values from the distribution of the radius r of the circle obtained by performing polar coordinate conversion on the detected circle with the coordinates existing in a fixed range as the origin, the coordinate of the origin where the evaluation value is highest is the center of the circle. And a circle center determining step for determining the position.

  According to the above invention, the Hough transform is performed on the image data, and the center position of the detected circle is temporarily set. Therefore, when a part of the circle of the image data is missing due to the influence of a foreign object or the like. Even in the case of image data in which uneven illumination (shading) appears, the center of the circle can be accurately obtained without being affected by noise by the Hough transformation.

  In the present invention, a predetermined range centered on the coordinates of the center position of the temporarily set circle is set, and the detected circle is subjected to polar coordinate conversion with each coordinate existing in the predetermined range as the origin. Then, the radius r of the circle obtained when the detected circle is converted into polar coordinates with each coordinate as the origin is projected one-dimensionally in the direction of the central angle θ axis.

  That is, by setting an origin candidate region for polar coordinate conversion centered on the center of the circle, performing polar coordinate conversion on all the regions, and performing a one-dimensional projection process, a polar coordinate system suitable for circle detection is used. , Image processing is performed with high accuracy, and polar coordinate conversion of the circle detected with all pixels existing in the set origin candidate area as the origin is performed, so the amount of data increases and the center of the circle is obtained with very high accuracy. Can do. Furthermore, by limiting the origin candidate area, the center position of the circle can be determined efficiently. Therefore, the processing time after the one-dimensional projection process can be shortened.

  Further, by performing the one-dimensional projection processing, an evaluation value indicating linearity with respect to the central angle θ axis direction of the radius r of the circle is obtained, so that the SN ratio can be increased. Furthermore, an optimal circle center can be selected from all origin candidates by a quantitative index called an evaluation value.

  As a result, when obtaining the circle center of the image data, it is possible to provide an image processing apparatus and an image processing method capable of accurately determining the center of the circle in a short time.

  In the image processing apparatus of the present invention, it is preferable that a smoothing processing means for performing a smoothing process on the image data before performing a Hough transform is provided.

  The image processing method of the present invention preferably includes a smoothing process step of performing a smoothing process on the image data before performing a Hough transform.

  As a result, the accuracy of the Hough transform can be improved by adding a smoothing process.

  In the image processing apparatus according to the present invention, the edge obtained in the radius r-axis direction is detected before the one-dimensional projection is performed on the circle obtained by the polar coordinate transformation, and binarization processing is performed with a predetermined value to perform edge processing. It is preferable that an edge processing means for extracting is provided.

  Further, in the image processing method of the present invention, the edge obtained in the radius r-axis direction is detected before one-dimensional projection is performed on the circle obtained by the polar coordinate transformation, and binarization processing is performed with a predetermined value to perform edge processing. It is preferable to include an edge processing step for extracting.

  As a result, binarization processing with a predetermined value is performed on a number of edge candidates that have been detected by edge detection, so that a clear edge, that is, an edge that has a specific strength and is somewhat strong is specified. be able to. Conversely, an unclear edge can be regarded as noise.

  In addition, the edge detection accuracy can be increased by performing, for example, a differentiation process after the polar coordinate conversion.

  In the image processing apparatus of the present invention, the evaluation value calculation means uses a threshold value from a distribution in the central angle θ-axis direction at a radius r of a circle obtained by one-dimensional projection by the one-dimensional projection means. It is preferable to obtain the evaluation value by a crystallization process.

  In the image processing method of the present invention, in the evaluation value calculation step, a threshold is used from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step. The evaluation value is preferably obtained by a valuation process.

  Thereby, after the one-dimensional projection, the calculation accuracy of the evaluation value indicating the linearity of the radius r of the circle with respect to the central angle θ-axis direction can be improved by the binarization process using the threshold value.

  In the image processing apparatus according to the present invention, the evaluation value calculation means may calculate a threshold value based on statistical data processing from a distribution in the central angle θ-axis direction at a radius r of a circle obtained by one-dimensional projection by the one-dimensional projection means. It is preferable to obtain the evaluation value using.

  In the image processing method of the present invention, in the evaluation value calculation step, statistical data processing is performed from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step. It is preferable to obtain the evaluation value using a threshold value.

  As a result, the threshold for binarization processing is obtained from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection process, such as the standard deviation of the one-dimensional image data. The threshold value can be determined dynamically by statistical data processing.

  That is, by dynamically determining, it is possible to follow when image data changes due to illumination conditions or the like.

  In the image processing apparatus according to the aspect of the invention, it is preferable that the evaluation value calculation unit uses an inverse of the sum of luminance values of image data equal to or less than the threshold as an evaluation value indicating the linearity.

  In the image processing method of the present invention, it is preferable that, in the evaluation value calculating step, an inverse number in the sum of luminance values of image data equal to or less than the threshold value is an evaluation value indicating the linearity.

  Thereby, the evaluation value which shows linearity can be obtained by simple processing with a high processing speed. In addition, for example, when the image is captured under different conditions such as an image with a part of a circle missing, an image with a foreign object on the circumference, a broken circle image, an image with uneven illumination, etc., the evaluation value of the present invention, It can be evaluated quantitatively.

  Further, in the image processing apparatus of the present invention, before the evaluation value is obtained with respect to a circle obtained by performing polar coordinate conversion on a circle detected using each coordinate existing in the predetermined range as an origin, the interruption is interrupted. It is preferable to provide expansion / contraction processing means for performing expansion processing for connecting the circles and performing contraction processing for removing noise from the circle when the polar coordinates are converted.

  Further, in the image processing method of the present invention, before obtaining an evaluation value for a circle obtained by performing polar coordinate conversion on a circle detected with each coordinate existing in the predetermined range as an origin, the interruption is interrupted. It is preferable to include an expansion / contraction process step of performing an expansion process for connecting the circles and performing a contraction process for removing noise from the circle when the polar coordinates are converted.

  Thus, by adding expansion / contraction processing with respect to the central angle θ-axis direction before obtaining the evaluation value, it is possible to connect a straight line interrupted by the expansion processing, and noise can be deleted by the contraction processing. The accuracy can be increased after the two-dimensional projection process.

  An image processing program of the present invention is an image processing program for causing a computer to function as the above-described image processing apparatus in order to solve the above-described problem, and causes the computer to function as each of the above-described means. It is said.

  According to the above invention, it is possible to provide an image processing program that can determine the center of a circle with high accuracy in a short time when determining the center of a circle of image data.

  The computer-readable recording medium of the present invention is characterized in that the above-described image processing program is recorded in order to solve the above-described problems.

  According to the above invention, the image processing apparatus can be realized on a computer such as an LSI capable of executing the program by hardware by the image processing program read from the recording medium.

  In order to solve the above problems, an image processing system according to the present invention generates image data by capturing an image of an object with an imaging device, and performs image processing on the image data. The position of the optical axis of the Fresnel lens is provided between the imaging device and a Fresnel lens made up of multiple concentric convex lenses, and the optical axis of the Fresnel lens is disposed on the imaging target position of the object. And calculating the position of the optical axis of the Fresnel lens by the image processing apparatus so that the optical axis of the Fresnel lens is arranged on an imaging target position of the object. And a drive means for moving the Fresnel lens.

  According to the above-described invention, when it is desired to collect parallel light on an object using a Fresnel lens having a concentric shape change on the circumference with the center of the lens as a circle, the center position of the Fresnel lens and the object The method for determining the center position of a circle according to the present invention can be applied to alignment of an object with an imaging target position.

  As described above, the image processing apparatus of the present invention performs a Hough transform on image data and temporarily sets the center position of the detected circle, and the center of the temporarily set circle. A predetermined range centered on the coordinates of the position is set, polar coordinate conversion means for converting the detected circle into a polar coordinate with each coordinate existing in the predetermined range as an origin, and the detected circle with each coordinate as an origin One-dimensional projection means for one-dimensionally projecting the circle radius r obtained by polar coordinate conversion in the direction of the central angle θ axis, respectively, and the central angle θ at the radius r of the circle obtained by one-dimensional projection by the one-dimensional projection means Evaluation value calculation means for obtaining an evaluation value indicating linearity of the radius r of the circle with respect to the central angle θ axis direction from the distribution in the axial direction, and the detected circle with the coordinates existing in the predetermined range as the origin as polar coordinates Yen obtained by conversion Among the evaluation values from the distribution of the radius r, in which the circle center determining means for determining the origin of coordinates evaluation value is highest as the center position of the circle is provided.

  In addition, as described above, the image processing method of the present invention performs a Hough transform on the image data, temporarily sets the center position of the detected circle, and the temporarily set circle. A polar range conversion step of setting a predetermined range centered on the coordinates of the center position of the center and converting the detected circle into a polar coordinate with each coordinate existing in the predetermined range as the origin, and the above-described detection with each coordinate as the origin A one-dimensional projection step for one-dimensionally projecting a circle radius r obtained by converting the circle into polar coordinates in the direction of the central angle θ axis, and a center at the radius r of the circle obtained by the one-dimensional projection by the one-dimensional projection step An evaluation value calculation step for obtaining an evaluation value indicating linearity of the radius r of the circle with respect to the central angle θ axis direction from the distribution in the angle θ axis direction, and the circle detected using the coordinates existing in the predetermined range as the origin Is obtained by transforming Among the evaluation values from the distribution of the radius r of the circle was, the method comprising the circle center determining step of determining the origin of coordinates evaluation value is highest as the center position of the circle.

  As described above, the image processing program of the present invention is an image processing program for causing a computer to function as the above-described image processing apparatus, and causes the computer to function as each of the above-described means.

  The computer-readable recording medium of the present invention records the above-described image processing program as described above.

  In addition, as described above, the image processing system of the present invention is provided with a Fresnel lens made up of multiple concentric convex lenses between the object and the imaging device, and the optical axis of the Fresnel lens is used to image the object. Based on the calculation result of the position of the optical axis of the Fresnel lens by the image processing apparatus and the image processing apparatus described above for calculating the position of the optical axis of the Fresnel lens so as to be placed on a target position, Drive means for moving the Fresnel lens is provided so that the optical axis of the Fresnel lens is disposed on the imaging target position of the object.

  Therefore, it is possible to provide an image processing apparatus, an image processing method, an image processing program, a recording medium, and an image processing system that can accurately determine the center of a circle in a short time when determining the circle center of image data. Play.

[Embodiment 1]
An embodiment of the present invention will be described below with reference to FIGS. 1 to 9 and FIG.

  The image processing apparatus according to the present embodiment performs graphic processing on image data input from an image input apparatus such as an imaging apparatus. Specifically, the center position of a circular image included in the image data is determined. Image processing to be performed. In the present invention, the image data is not necessarily input from an image input device such as an imaging device.

  An image processing system 1a according to the present embodiment including the image processing device includes an imaging device 2 and an image processing system main body 10 as shown in FIG.

  The imaging device 2 is composed of, for example, an image input device such as a CCD (Charge Coupled Device) camera, and the captured image data is transferred to the image processing system main body 10 by imaging the object 3. ing. Note that the imaging device 2 is not necessarily limited to a CCD camera, and may be a camera including another imaging sensor, and is not limited to a camera, and may be an image input device such as a scanner. The acquisition source of the image data is not necessarily limited to the imaging device 2 and may be a database in which image data is stored in advance.

  In the present embodiment, the image processing system 1a captures products and parts with a camera in order to realize, for example, automation of product assembly, automation of inspection work, position adjustment of product assembly, and the like with high accuracy. In addition, since various image processing is performed on the obtained image data, the object 3 is a product and a part used for product assembly. However, the present invention is not necessarily limited thereto, and in the present invention, the object 3 in another field may be used.

  The image processing system main body 10 includes an image processing device 20, a control device 11, a recording device 12, and an output device 13.

  The control device 11, the recording device 12, and the output device 13 have functions described in Embodiment 3 to be described later, in addition to those related to general control of the image processing system main body 10.

  As shown in FIG. 3, the image processing apparatus 20 includes a circle center temporary setting unit 21 as a circle center determination unit, a polar coordinate conversion unit 22 as a polar coordinate conversion unit, and a one-dimensional projection unit 23 as a one-dimensional projection unit. And an evaluation value calculation unit 24 as an evaluation value calculation unit, and a circle center determination unit 25 as a circle center temporary setting unit.

  The circle center temporary setting unit 21 performs Hough conversion on the image data captured by the imaging device 2 to temporarily set the center position of the circle. Further, the polar coordinate conversion unit 22 performs polar coordinate conversion using each coordinate in a predetermined range centered on the center position of the circle temporarily set by the circle center temporary setting unit 21 as the origin. The one-dimensional projection unit 23 projects one-dimensionally in the θ-axis direction for each image data obtained by the polar coordinate conversion. Furthermore, the evaluation value calculation unit 24 obtains an evaluation value indicating linearity from each one-dimensionally projected data. Then, the circle center determining unit 25 determines the coordinate having the highest evaluation value among the coordinates in the predetermined range as the center position of the circle.

  An image processing method by the image processing apparatus 20 having the above configuration will be described with reference to FIGS. FIG. 1 is a flowchart illustrating an image processing method for determining the center position of a circle according to the present embodiment, and FIG. 4 is a diagram illustrating image data captured by the imaging apparatus 2.

  Here, as shown in FIG. 4, in the present embodiment, it is assumed that the image data has, for example, uneven illumination and the circle is interrupted. Note that the unevenness of the image data is not necessarily limited to this, and for example, there are also those shown in FIGS. That is, as shown in FIG. 23A, a part of the circle is missing, or as shown in FIG. 23B, foreign matter, scratches, noise, or the like F1 exists on the circumference, or FIG. ), As shown in FIG. 23 (d), the unevenness of the brightness occurs due to the circle being discontinuous or due to the relative positional relationship between the illumination and the casing or parts. This is a case where part of the image data is missing. These are merely examples, and the present invention is not necessarily limited thereto. In this embodiment, the present invention can be applied to these image data.

  In the flowchart shown in FIG. 1, first, image data is input (S1). Next, the image data is subjected to Hough transform to detect a circle (S2). In detecting a circle, when a plurality of circles are detected, the circle may be limited to a specific circle, or the processing after S3 may be performed for all the circles to determine the center positions of the plurality of circles. .

  Here, the Hough transform is a P.V. V. C. This is a technique proposed by Hough, and is used to detect straight lines and circles. In the case of detecting a straight line, a point (x, y) on the original rectangular coordinate is converted into a polar coordinate two-dimensional space having a central angle θ and a distance ρ, and the number of the points is stored in the memory array for each central angle θ and the distance ρ. Add to. Then, the combination of the center angle θ and the distance ρ, which has the largest number, is returned to the original rectangular coordinates to form a collection of points that are most likely to be straight. If the number is lowered, the next candidate is obtained sequentially. If the central angle θ and the distance ρ are subdivided, the accuracy increases. In the case of detecting a circle, the original point (x, y) on the rectangular coordinates is converted into a three-dimensional space of the circle center point (X, Y) and the radius r, and the same processing is performed.

  The principle of straight line detection or circle detection by this Hough transform is to count the number of points on a virtual straight line or circumference, and if there are many, it is regarded as a straight line or circle, so the points are randomly If there are many, a plurality of straight lines or circles may be detected.

  A circle detection method based on the Hough transform will be described with reference to FIG. FIG. 5 is an explanatory diagram showing a circle detection method.

As shown in FIG. 5, all the circles passing through the point (x, y) on the Cartesian coordinates are represented by the center point (center X, center Y) and radius (r) of the circle. The Hough transform changes the center X and center Y,
r ² = (x−center X) ² + (y−center Y) ²
From the relationship, the radius r is obtained.

  Using the above equation, when a point (x, y) on a rectangular coordinate is transformed onto a new three-dimensional space (center X, center Y, r), one point on the rectangular coordinate is a plane in the three-dimensional space. Corresponding to If there are many points on the rectangular coordinates, a large number of curved surfaces can be obtained in space. If there are points shared by these curved surfaces, they will be arranged on a circle on the original xy rectangular coordinates.

  In this way, as shown in S2 of FIG. 1, a circle is detected by Hough transform.

  Next, the center position coordinates of the circle are obtained (S3), and a predetermined range having a predetermined width centered on the obtained center position coordinates is set as a range of candidate points for the center position. Then, the polar coordinates of the detected circle are converted from the coordinates existing within the predetermined range with the center position coordinate P as the origin (S5).

  That is, as shown in FIG. 6A, an origin candidate region D that is a predetermined range having a predetermined width centered on the center position coordinate P is set. And as a coordinate inside the origin candidate area | region D, when a center position coordinate is set to P (i, j), as shown in FIG.6 (b), P (i + n, j + m) (n = -2-2). , M = −2 to 2) are the coordinates of the candidate points.

  As described above, the origin candidate region D as the predetermined range is set, for example, with respect to the orthogonal coordinate system, for example, within n = ± 2 pixels in the x direction of the center position coordinate P of the circle, and m = ± in the y direction. It is set in advance such that it is within 2 pixels. About the value of these settings, it can carry out by extraction of the range with high presence frequency by statistical data.

  Note that the origin candidate region D as the predetermined range is not necessarily rectangular, and may be a rhombus or a circle. Further, n = ± 2 pixels or less and m = ± 2 pixels or less are examples, and are not necessarily limited thereto. Furthermore, although the orthogonal coordinate system is adopted as the origin candidate region D, the present invention is not necessarily limited to this, and a polar coordinate system can also be adopted. When the polar coordinate system is adopted, the radius r is set to a predetermined range.

  Further, in the present embodiment, as coordinates existing inside the origin candidate region D, P (i + n, j + m) (n = −2 to 2, m = −2 to 2) are arranged in a grid in order. Is adopted. And thereby, all the coordinates which exist in the origin candidate area | region D are employ | adopted as a candidate point. As a result, the center of the circle can be obtained with higher accuracy. In addition, when there are few candidate points, there is not always a case in which the set corresponds to the true circle center. On the other hand, in this embodiment, since the circumference of the circle center candidate is thoroughly examined, a true circle center with higher accuracy can be obtained.

  The selection of the coordinates existing inside the origin candidate area D is not necessarily limited to this in the present invention, and every other pixel or other random coordinates can be adopted. Also by this, for example, the inside of a wide range of origin candidate regions D can be uniformly investigated.

  In S5 of FIG. 1, the polar coordinate conversion means that the orthogonal coordinate system is converted to a polar coordinate system of r-axis and θ-axis as shown in FIG.

  Next, one-dimensional projection processing and normalization are performed with respect to the θ-axis direction (S6). Normalization is desirably divided by the number of projected data, for example.

  Next, an evaluation value is calculated (S7). As shown in FIGS. 7A and 7B and FIGS. 8A and 8B, this evaluation value is the reciprocal of the sum of the one-dimensional data that is equal to or less than the threshold value TH. That is, the evaluation value is a value indicating linearity with respect to the central angle θ-axis direction of the radius r of the circle from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by one-dimensional projection. In the present embodiment, as shown in FIG. 9A, when the area of the hatched portion below the threshold TH is S1, the evaluation value = 1 / S1. In this way, when the radius r is substantially a straight line with respect to the central angle θ-axis direction as shown in FIG. 8A by using the reciprocal of the sum of the one-dimensional data, FIG. As shown, the one-dimensional projection data in the direction of the central angle θ axis of the radius r of the circle becomes sharp, and as a result, the evaluation value increases.

  Note that the evaluation value only needs to be a profile with a narrow peak as shown in FIG. 8B and a high evaluation value as a high evaluation value. Therefore, as shown in FIG. 9B, the half-value width W × peak value H May be used. However, considering that processing time is required, as described above, it is desirable to reciprocate the sum of the one-dimensional data that is equal to or less than the threshold value TH. Further, as shown in FIG. 9C, in the present embodiment, the threshold value TH is preferably set to three times the standard deviation σ of the one-dimensional data and dynamically changed. This is because if the threshold value is a static threshold value, the threshold value TH needs to be reset according to the situation. That is, the variation tends to increase when the image is bright, and the variation decreases when the image is dark. Therefore, by setting 3σ as a threshold value, an accurate radius r can be obtained whether the image is bright or the image is dark. On the other hand, if the threshold value TH is fixed, the brightness of the image cannot be handled.

  Next, circles obtained by one-dimensional projection in the central angle θ-axis direction with respect to the radius r of the circle obtained when the coordinates of the detected circle set in S5 shown in FIG. Whether or not all evaluation values indicating linearity with respect to the central angle θ axis direction of the radius r of the circle are calculated from the distribution in the central angle θ axis direction at the radius r of the circle (S8), and candidate points remain. Then, the process returns to S5. In this way, the processes in S5 to S8 are repeated for all candidate points set in S5.

  If it is determined in S8 that all the evaluation values set in S5 have been calculated, the candidate point with the highest evaluation value is determined as the center position of the circle (S9). In this flowchart, since one-dimensional data is handled particularly after S6, the processing time can be shortened.

  As described above, the image processing apparatus 20 and the image processing method of the present embodiment are the image processing apparatus 20 and the image processing method for obtaining the center position of the circle displayed by the image data. Then, a Hough transformation is performed on the image data, a circle center temporary setting step for temporarily setting the center position of the detected circle, and a predetermined range centered on the coordinates of the center position of the temporarily set circle are set. A polar coordinate conversion step of converting the detected circle with the coordinates as the origin in the predetermined range, and a radius r of the circle obtained when the detected circle is converted into the polar coordinates with the coordinates as the origin. From the one-dimensional projection step for one-dimensional projection in the central angle θ-axis direction and the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection by the one-dimensional projection step, the radius r of the circle From an evaluation value calculation step for obtaining an evaluation value indicating linearity with respect to the central angle θ-axis direction, and a distribution of radius r of the circle obtained by performing polar coordinate conversion on the detected circle with each coordinate existing in the predetermined range as the origin Of each evaluation value of Evaluation value contains a circle center determining step of determining the highest becomes the origin of the coordinate as the center position of the circle.

  As a result, the Hough transformation is performed on the image data, and the center position of the detected circle is temporarily set. Therefore, when a part of the circle of the image data is missing due to the influence of a foreign object or the like, Even in the case of image data in which unevenness (shading) appears, the center of the circle can be obtained without being influenced by noise by the Hough transform. In addition, since a circle is detected by Hough transformation based on polar coordinates, for example, it is more accurate than a method in which an arbitrary two points on a circular arc are selected and a vertical bisector is drawn to obtain a center. Can be sought. That is, the method of selecting an arbitrary two points in a point sequence on an arc and drawing a vertical bisector to obtain the center is strongly influenced by noise. In order to avoid this effect, it is not sufficient to find the most probable center by increasing the number of samples by an appropriate number.

  In the present embodiment, a predetermined range centered on the coordinates of the center position of the temporarily set circle is set, and the detected circle is subjected to polar coordinate conversion with each coordinate existing in the predetermined range as the origin. Then, the radius r of the circle obtained when the detected circle is converted into polar coordinates with each coordinate as the origin is projected one-dimensionally in the direction of the central angle θ axis.

  That is, an origin candidate region D for polar coordinate conversion centered on the center of a circle is set, polar coordinate conversion is performed for all the regions, and a one-dimensional projection process is performed, thereby using a polar coordinate system suitable for circle detection. Thus, the image processing is performed with high accuracy, and the polar coordinate conversion of the circle detected using all the pixels existing in the set origin candidate region D as the origin is performed. Can be sought. Furthermore, by limiting the origin candidate region D, the center position of the circle can be determined efficiently. Therefore, the processing time after the one-dimensional projection process can be shortened.

  Further, by performing the one-dimensional projection processing, an evaluation value indicating linearity with respect to the central angle θ axis direction of the radius r of the circle is obtained, so that the SN ratio can be increased. Furthermore, an optimal circle center can be selected from all origin candidates by a quantitative index called an evaluation value.

  As a result, it is possible to provide the image processing apparatus 20 and the image processing method that can determine the center of the circle with high accuracy and in a short time when determining the center of the circle of the image data.

  Further, in the image processing apparatus 20 and the image processing method of the present embodiment, in the evaluation value calculation step, from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step. The evaluation value is obtained by binarization processing using the threshold TH.

  Thereby, after one-dimensional projection, the calculation accuracy of the evaluation value indicating the linearity of the radius r of the circle with respect to the central angle θ-axis direction can be increased by the binarization process using the threshold value TH.

  In the image processing apparatus 20 and the image processing according to the present embodiment, in the evaluation value calculation step, the statistical value is calculated from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step. An evaluation value is obtained by using a threshold value TH by automatic data processing.

  As a result, the threshold TH for binarization processing is determined from, for example, the standard deviation σ of the one-dimensional image data from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step. The threshold value TH can be determined dynamically by statistical data processing such as

  That is, by dynamically determining, it is possible to follow when image data changes due to illumination conditions or the like. Further, in the present embodiment, since the comparison point is quantitatively determined for the candidate point set from the provisional circle center, the accuracy of setting the circle center is high.

  In the image processing device 20 and the image processing according to the present embodiment, in the evaluation value calculation step, the reciprocal of the sum of the luminance values of the image data equal to or lower than the threshold value TH is used as an evaluation value indicating linearity. In the present embodiment, the brightness value of the image data is adopted, but the present invention is not necessarily limited to the brightness value, and may be a density, a gradation value, a brightness value, a shading value, or the like of the image data.

  Thereby, the evaluation value which shows linearity can be obtained by simple processing with a high processing speed. In addition, for example, when the image is captured under different conditions such as an image with a part of a circle missing, an image with a foreign object on the circumference, a broken circle image, an image with uneven illumination, etc., the evaluation value of the present invention, It can be evaluated quantitatively.

  Further, in the image processing apparatus 20 and the image processing according to the present embodiment, before the evaluation value is obtained for the circle obtained by performing polar coordinate conversion on the circle detected using each coordinate existing in the predetermined range as the origin, the interruption is performed. It includes an expansion / contraction process step of performing expansion processing to connect the broken circles and performing contraction processing to delete noise from the circle when the polar coordinate conversion is performed.

  Thus, by adding expansion / contraction processing with respect to the central angle θ-axis direction before obtaining the evaluation value, it is possible to connect a straight line interrupted by the expansion processing, and noise can be deleted by the contraction processing. The accuracy can be increased after the two-dimensional projection process.

[Embodiment 2]
The following will describe another embodiment of the present invention with reference to FIGS. Configurations other than those described in the present embodiment are the same as those in the first embodiment. For convenience of explanation, members having the same functions as those shown in the drawings of the first embodiment are given the same reference numerals, and explanation thereof is omitted.

  In addition to the configuration of the image processing apparatus 20 of the first embodiment, the image processing apparatus 30 of the present embodiment has a smoothing processing unit 31 as a smoothing processing unit and an edge processing unit as shown in FIG. Edge processing section 32 and expansion / contraction processing section 33 as expansion / contraction processing means.

  The smoothing processing unit 31 performs a smoothing process on the image data before performing the Hough transform. The edge processing unit 32 performs edge processing in the r-axis direction before the one-dimensional projection process in the θ-axis direction. Furthermore, the expansion / contraction processing unit 33 performs expansion / contraction processing of a predetermined size in the θ-axis direction.

  An image processing method by the image processing apparatus 30 having the above configuration will be described with reference to FIG. FIG. 11 is a flowchart showing an image processing method for determining the center position of a circle according to the present embodiment. The same steps as those in FIG. 1 are given the same step numbers. The image data is the same as the image data shown in FIG.

  As shown in FIG. 11, image data is input (S1). Next, smoothing processing is performed on the image data.

  Here, the smoothing process will be described in detail. Here, the case where the pixel values of the original image in the simplest 3 × 3 pixel range are averaged and output as the central pixel value will be described. This smoothing process is a kind of operation for taking a moving average, and the image is “blurred” by being smoothed.

  For example, assume that there is unprocessed 3 × 3 pixel image data as shown in FIG. When the value of the center pixel in the unprocessed 3 × 3 pixel image data is “1” and the other pixels are “0”, smoothing processing is performed using averaging as a smoothing operator. Then, for example, it can be as shown in FIG. Further, as shown in FIG. 12C, by giving a large specific gravity to the value of the center pixel, the degree of “blur” can be made weaker than that in FIG. In each case, the sum of each element is 1.

  As described above, the smoothing process smoothes the center pixel when the center pixel is different from the gradation or luminance of the surrounding pixels. Thereby, for example, when the noise has a value different from that of the surrounding pixels, the noise can be smoothed. Then, by using various smoothing operators, it is possible to make the degree of smoothing appropriate. In order to make the degree of smoothness appropriate, for example, the range of “blur” can be adjusted by σ (sigma) using a Gaussian distribution. When a Gaussian distribution is used, for example, smoothing is performed using a matrix filter of 7 × 7 to 15 × 15 pixels.

  In the present embodiment, smoothing is performed using, for example, a Gaussian distribution by a 9 × 9 pixel Gaussian filter due to the combination with the Hough transform. When smoothing using a Gaussian distribution by a Gaussian filter, when the image is I and the Gaussian distribution is G (σ), the smoothed image L (σ) is calculated by the following equation.

L (σ) = G (σ) * I
In this embodiment, as shown in FIG. 11, after performing the above smoothing process in S11, Hough transform is performed to detect a circle (S2). As described above, when a plurality of circles are detected, the radius may be limited to a specific circle, or the processing after S3 may be performed for all the circles to determine the center positions of the plurality of circles. .

  Next, the center coordinates of the circle are obtained (S3). As shown in FIGS. 6 (a) and 6 (b), an origin candidate area that is a candidate point area having a predetermined width centered on the obtained center position coordinates P. Set D. This origin candidate area D indicates an area where the true center position coordinate P may exist.

  Next, after performing polar coordinate conversion with the center position coordinate P, which is the obtained candidate point, as the origin (S5), in this embodiment, edge detection in the radius r-axis direction is performed (S12). For example, an edge can be detected by r-direction Sobel filtering. As described above, in the present embodiment, since edge detection in the radius r-axis direction is performed after polar coordinate conversion, it is difficult to be affected by noise such as foreign matter. That is, in an orthogonal coordinate system, when an edge is a curve, edge detection cannot be performed successfully. On the other hand, in the present embodiment, the detection target is assumed to be a circle, and the influence of noise can be minimized by detecting the edge in the radius r-axis direction after polar coordinates.

  Here, edge detection by Sobel filter processing will be described in detail.

  First, there are edge detection operators that use primary differentiation and those that use secondary differentiation. The first derivative has directionality and needs to be detected separately in the vertical direction, the horizontal direction, and the like. Among the famous first-order differential methods are Prewitt and Sobel, but there is little difference between the two.

  On the other hand, Laplacian is used for the second derivative. The second derivative using Laplacian is not directional but has high sensitivity, so it has a weakness to noise. Therefore, this is rarely used alone and often used together with Gaussian blurring (smoothing).

  In the present embodiment, a general Sobel operator is used in the first-order differentiation. The Sobel operator uses 3 × 3 matrix edge detection filters, each of which is a vertical edge detection filter shown in FIG. 13 (a) and a horizontal edge detection filter shown in FIG. 13 (b). This filter is a filter that calculates a difference in luminance between adjacent pixels in the vertical direction and the horizontal direction as a first derivative. In the edge portion of the image, the absolute value of this difference becomes a large value, and this is based on the principle that the edge can be detected. In other words, in the edge portion of the image, the color of adjacent pixels changes abruptly, for example, from black to white and from white to black, and the difference in luminance increases.

  Specifically, for example, a method for detecting an edge using a vertical edge detection filter will be described with reference to FIGS. FIGS. 14A to 14D conceptually show the outputs when the vertical edge detection filter and various images are multiplied, and FIGS. 14A to 14D. The 3 × 3 matrix on the left side of each is a vertical edge detection filter, and the 3 × 3 matrix on each right side is actual image data. Here, for simplification of explanation, the luminance value of the black portion is 0, and the luminance value of the white portion is 1.

  That is, as shown in FIG. 14A, when a 3 × 3 vertical edge detection filter is multiplied by a 3 × 3 image having a black portion on the right side, an output of “4” is obtained. As shown in (b), when a 3 × 3 vertical edge detection filter is multiplied by a 3 × 3 image having a black portion on the left side, “−4” is obtained as an output. On the other hand, as shown in FIG. 14C, when there are black portions above and below, or when there is no change in the black portion or white portion as shown in FIG. 14D, the edge detection result is “0”. become. As a result, when the absolute value of the output is large, it means that an edge exists in that portion. Further, as shown in FIGS. 14 (a) and 14 (b), the output becomes large when there is a black portion in the vertical direction, so that it is understood that this filter can detect the edge in the vertical direction. Further, whether the right side is black or the left side is black can be recognized by the sign of the output. In this embodiment, the direction of the edge is not a problem and is handled as an absolute value.

  As described above, the edge in the radius r direction is detected as shown in S12 of FIG. In the above description, the description has been made in the orthogonal coordinate system such as the vertical direction and the horizontal direction. However, in the present embodiment, for example, the horizontal direction is the radial r-axis direction, and the vertical direction is perpendicular to the radial r-axis direction. Edge detection is performed by replacing the correct direction with the vertical direction.

  Next, binarization processing is performed with a threshold value as a predetermined value (S13), and the threshold value is limited to those with strong edges. That is, in the example of FIGS. 14A to 14D described above, for example, when the threshold is “3”, an output that is “3” or more is extracted as an edge. Specifically, FIG. 4A and FIG. 4B are extracted as edges.

  Next, expansion / contraction processing is performed in the θ-axis direction (S14). Thereby, as shown in FIGS. 15 (a) and 15 (b), it is possible to connect the broken circles by the expansion process, and also to connect the broken circles in FIGS. 16 (a) and 16 (b). it can. Note that when noise other than the direction of the central angle θ-axis exists in the circle when the polar coordinates are converted, the noise can be deleted.

  Next, one-dimensional projection processing and normalization are performed with respect to the central angle θ-axis direction (S6). For example, normalization is preferably divided by the number of projected data. As a result, FIG. 7B and FIG. 8B are obtained.

  Next, an evaluation value is calculated (S7). As shown in FIGS. 7A and 7B and FIGS. 8A and 8B, this evaluation value is the reciprocal of the sum of the one-dimensional data that is equal to or less than the threshold value TH. Since this evaluation value only needs to show a high profile with a thin peak and a high profile as shown in FIG. 8 (b), half-value width × peak value may be used, but processing time is required. In consideration of the above, as described above, it is desirable to reciprocate the sum of the one-dimensional data that is equal to or less than the threshold value TH. Further, the threshold TH here is preferably a threshold that is three times the standard deviation σ of the one-dimensional data and dynamically changes. This is because if the threshold value is a static threshold value, the threshold value TH needs to be reset according to the situation.

  Then, it is determined whether all the evaluation values set in S5 have been calculated (S8), and if candidate points remain, the process returns to S5. In this way, the processes in S5 to S8 are repeated for all candidate points set in S5.

  If it is determined in S8 that all the evaluation values set in S5 have been calculated, the candidate point with the highest evaluation value is determined as the center position of the circle (S9). In this flowchart, since one-dimensional data is handled particularly after S6, the processing time can be shortened.

  As described above, the image processing apparatus 30 and the image processing method according to the present embodiment include a smoothing process step of performing a smoothing process on the image data before performing the Hough transform.

  As a result, the accuracy of the Hough transform can be improved by adding a smoothing process.

  Further, the image processing apparatus 30 and the image processing method according to the present embodiment include an edge processing step of extracting edges in the radius r-axis direction before one-dimensional projection with respect to a circle obtained by polar coordinate conversion.

  Thereby, edge detection accuracy can be improved by performing a differentiation process after polar coordinate conversion.

[Embodiment 3]
The following will describe another embodiment of the present invention with reference to FIGS. Note that configurations other than those described in the present embodiment are the same as those in the first embodiment and the first embodiment. For convenience of explanation, members having the same functions as those shown in the drawings of Embodiment 1 and Embodiment 2 described above are given the same reference numerals, and descriptions thereof are omitted.

  As shown in FIG. 17, the image processing system 1 b according to the present embodiment is provided with a sheet-like Fresnel lens 4 in which multiple concentric circles that are condensing lenses are formed between the imaging device 2 and the object 3. Is possible. The imaging device 2 can magnify an image of the object 3 by passing it through the Fresnel lens 4.

  That is, the Fresnel lens 4 is provided with a driving device 5 as a driving means, and the retracted Fresnel lens 4 is translated horizontally by the driving device 5 so that the imaging device 2 and the object 3 are moved. It can be inserted in between. The driving device 5 drives the Fresnel lens 4 by gripping the Fresnel lens 4 and its holding member (not shown) from the side, or supports the periphery of the Fresnel lens 4 and the holding member so as to be horizontally movable. It is preferable that Moreover, it is preferable that the drive device 5 further has a tilt angle adjustment in order to arrange the Fresnel lens 4 in parallel.

  By the way, when the Fresnel lens 4 is inserted between the imaging device 2 and the object 3, the imaging target position of the object 3 on which the parallel light is to be collected and the center of the Fresnel lens 4 are installed on the same axis. There is a need to.

  However, the Fresnel lens 4 is a lens whose shape has changed along the circumference on the surface. As shown in FIG. 18, a plurality of concentric circles appear, but the illumination is not uniformly irradiated to the Fresnel lens 4. In some cases, you may be interrupted. Therefore, it is necessary to find the center of the circle from this concentric circle.

  That is, as a basic idea, in the present embodiment, the Fresnel lens 4 is arranged so that the center of the Fresnel lens 4 and the target imaging position of the object 3 are the same position (the x coordinate and the y coordinate coincide). The purpose is to move and install.

  For example, there is an XYZ stage as means for moving the Fresnel lens 4, and the technique described in the first and second embodiments is used as means for obtaining the center coordinates of the Fresnel lens 4. Since the object 3 is fixed and does not move, the image is once acquired by the imaging device 2 and the coordinates are obtained by using a known technique or the like, or the person inputs the coordinates based on the image, etc. Find the coordinates of the object 3.

In particular,
(1) The position coordinates of the object 3 are obtained (x0, y0).

  (2) The Fresnel lens 4 is inserted and imaging is performed.

  (3) The center position of the Fresnel lens 4 is obtained (x1, y1).

(4) The amount of movement (x0-x1, y0-y1) of the XYZ stage is determined from the positional deviation between the center position of the Fresnel lens 4 and the target imaging position of the object 3, and the XYZ stage is moved. In some cases, imaging is performed again, and (3) to (4) are repeated.
In this procedure, the Fresnel lens 4 is moved and fixed so that the center of the Fresnel lens 4 and the imaging target position of the object 3 are the same position (the x coordinate and the y coordinate coincide). .

  A method of setting the imaging target position of the object 3 and the center of the Fresnel lens 4 on the same axis in the image processing system 1b having the above configuration will be described in detail below.

  First, the Fresnel lens 4 is moved to a position where the Fresnel lens 4 does not enter the field of view of the imaging device 2. Then, the object 3 is imaged by the imaging device 2, the coordinates of the portion where the object 3 exists are obtained by the image processing devices 20 and 30, and stored in the recording device 12. The coordinates of the object 3 here can be easily obtained by using, for example, simple binary image processing.

  Next, the center position of the Fresnel lens 4 is obtained using the imaging apparatus 2 and the image processing apparatuses 20 and 30 using the circle center position determination method described in the first or second embodiment. Then, the position difference (position difference amount) between the obtained center position and the position of the object 3 stored in the recording device 12 is obtained.

  Based on this positional difference, by inserting the Fresnel lens 4 with the driving device 5, the center of the object 3 on which the parallel light is to be collected and the center of the Fresnel lens 4 can be placed on the same axis. .

  Note that the position of the Fresnel lens 4 may be automatically adjusted by moving the drive device 5 with the control device 11, or a person manually operates based on the value (amount of position difference) displayed on the output device 13. Thus, the position of the Fresnel lens 4 may be adjusted.

  A method for determining the center position of the concentric circles in the Fresnel lens 4 will be described in detail below.

  As in the first and second embodiments, a two-stage process is performed. That is, in the first coarse search, the center of the circle used for the polar coordinate transformation is obtained by the Hough transformation, and in the second dense search, the polar coordinate transformation is performed, the edge is detected, and the circle center is specified.

  First, the first coarse search will be described based on the flowchart shown in FIG. 19 and FIG.

  As shown in FIG. 19, first, smoothing is performed as preprocessing (S21). In this case, for example, a 9 × 9 matrix Gaussian filter is used. Next, circle detection by Hough transformation is performed (S22), and circle candidate groups are narrowed down (S23). That is, the narrowing of the circle candidate group is processing when a plurality of circles are detected, and in the present embodiment, the center of the Fresnel lens 4 regarding the concentric circle is obtained, so that a plurality of circles are detected.

  Thereafter, the center coordinates of the circle are calculated (S24). Thereby, as shown in FIG. 20, the center position detected by the rough search is obtained.

  The important setting parameters in this rough search are the smoothing size, the minimum distance between detection circles that are parameters of the Hough transform (decreasing the number of false detections), the threshold for edge detection, and the center of the circle Detection threshold, circle allowable radius, and center range allowable range, which are conditions for selecting a detection circle candidate. Note that these setting parameters are examples and may be different.

  Next, the second dense search will be described based on the flowchart shown in FIG. 21 and FIGS. 22 (a) and 22 (b).

  First, as shown in FIG. 21, one of the coordinate groups obtained by the coarse search is set as the polar coordinate conversion center (S31), and polar coordinate conversion is performed (S32). Thereby, FIGS. 22A and 22B are obtained. That is, when away from the center of the circle, as shown in FIG. 22A, the radius r undulates along the central angle θ-axis direction as in FIG. 7A described above. On the other hand, in the case of the vicinity of the center of the circle, as shown in FIG. 22 (b), the radius r becomes linear along with the central angle θ-axis as in FIG. 8 (a). Yes.

  Next, edge (Y direction) detection is performed (S33). This edge (Y direction) detection is performed by a Sobel filter process.

  Next, binarization is performed with threshold = 3σ (S34). Thereby, a strong edge can be left and noise can be removed. The image data is not information of “0” and “1”, and is data of 256 gradations (0-255) for an 8-bit image or 1024 gradations (0-1023) for a 10-bit image. Yes. Therefore, unlike the edge processing calculation shown in FIG. 14, a simple threshold cannot be used, and here, binarization is performed using threshold = 3σ. Here, σ is a standard deviation.

  Next, expansion / contraction processing, that is, processing for connecting lines is performed (S34).

  Next, one-dimensional (X direction) projection and normalization are performed (S36), projection component threshold processing is performed (S36), and integrated value calculation (evaluation value) is performed (S38). From this, the sum of the values below the threshold TH is normalized to obtain an evaluation value.

  Next, it is determined whether or not all the coordinate groups obtained by the coarse search have been set as the polar coordinate conversion center (S39). If one of the coordinate groups obtained by the coarse search remains, the polar coordinate conversion center coordinates are updated. After (S40), the process returns to S33 again. Then, S33 to S40 are repeated until all of the coordinate groups obtained by the coarse search are completed, and when it is determined that all of the coordinate groups obtained by the coarse search are finished, the polar coordinate conversion center coordinates having the smallest evaluation value are obtained. It is determined that the solution is a dense search solution (S41).

  In the flowchart of FIG. 21, parameters to be set include, for example, search area range / resolution [pix], binarization threshold, number of expansion / contraction processes, number of edge thinning, threshold after one-dimensional projection It is.

  As described above, in the image processing system 1b according to the present embodiment, image data is generated by imaging the object 3 with the imaging device 2, and image processing is performed on the image data. A Fresnel lens 4 composed of a convex lens having multiple concentric circles is provided between the object 3 and the imaging device 2, and the optical axis of the Fresnel lens 4 is arranged on the imaging target position of the object 3. In addition, based on the image processing devices 20 and 30 that calculate the position of the optical axis of the Fresnel lens 4 and the calculation result of the position of the optical axis of the Fresnel lens 4 by the image processing devices 20 and 30, the optical axis of the Fresnel lens 4 is A driving device 5 that moves the Fresnel lens 4 so as to be placed on the target imaging position of the object 3 is provided.

  Accordingly, when it is desired to collect parallel light on the object 3 using the Fresnel lens 4 having a concentric shape change on the circumference with the center of the lens as a circle, the center position of the Fresnel lens 4 and the object The method of determining the center position of the circle of the present embodiment can be applied to the alignment with the imaging target position 3.

  By the way, the image processing apparatuses 20 and 30 described in the first to third embodiments are arithmetic units that execute instructions of a program (control program, recording condition setting program) that realizes each function of the image processing apparatuses 20 and 30. A CPU (Central Processing Unit), a ROM (Lead Only Memory) as a storage means storing the program, a RAM (Random Access Memory) as a storage means for expanding the program, a storage means for storing the program and various data And a storage device (storage medium) (not shown) such as a memory.

  An object of the present invention is to provide a recording medium in which a program code (execution format program, intermediate code program, source program) of a control program, which is software that realizes the functions described above, can be read by a computer, the image processing apparatus This can also be achieved by supplying the program code 20 to 30 and reading and executing the program code recorded on the recording medium by the computer (or CPU or MPU). In this case, since the program code itself read from the recording medium realizes the above-described function, the recording medium on which the program code is recorded constitutes the present invention.

  Accordingly, it is possible to provide an image processing program and a recording medium that can accurately determine the center of a circle in a short time when determining the center of the image data.

  The present invention relates to an image processing device, an image processing method, an image processing program, a recording medium, and an image processing system that perform graphic processing on image data input from an image input device such as an imaging device. The present invention is applicable to an image processing apparatus, an image processing method, an image processing program, a recording medium, and an image processing system that determine the center position of a circular image included in image data.

1 is a basic flowchart illustrating a circle center position determination method according to an embodiment of an image processing method and an image processing apparatus of the present invention. It is a block diagram which shows the whole structure of the image processing system provided with the said image processing apparatus. It is a block diagram which shows the structure of the said image processing apparatus. It is a figure which shows the image data taken in from the imaging device at the time of image processing with the said image processing apparatus. It is a figure which shows the detection principle of the circle by Hough (Hough) conversion at the time of image processing with the said image processing apparatus. (A) is a figure which shows the candidate point area | region which is a possibility area | region where a true center position coordinate exists, when detecting a circle by the said Hough transformation and calculating | requiring the center, (b) It is an enlarged view which shows a candidate point area | region. (A) is a graph showing the result of polar coordinate conversion as a locus on the θ axis and the radius r axis when the variation of the radius r in each discontinuous circle read from the image data is large, and (b) is normalized It is a graph which shows distribution of the radius r of each circle after doing. (A) is a graph showing, as a locus on the θ-axis and the radius r-axis, the result of polar coordinate conversion when the variation of the radius r in each discontinuous circle read from the image data is small, and (b) is a normalization It is a graph which shows distribution of the radius r of each circle after doing. FIG. 7 is a block diagram illustrating a configuration of an image processing apparatus according to another embodiment of the image processing method and the image processing apparatus of the present invention. FIG. 7 is a block diagram illustrating a configuration of an image processing apparatus according to another embodiment of the image processing method and the image processing apparatus of the present invention. It is a flowchart which shows the circle center position determination method in the said image processing apparatus. (A) (b) (c) is a figure which shows the principle of the smoothing process in the said image processing apparatus. (A) is a figure which shows the filter for a vertical edge detection of the Sobel operator used by the edge detection in the said image processing apparatus, (b) is used by the edge detection in the said image processing apparatus. It is a figure which shows the filter for a horizontal edge detection of a Sobel operator. (A)-(d) is a figure which shows the principle of the edge detection by multiplying the said vertical direction edge detection filter and various images, and calculating | requiring an output. (A) is a graph showing, as a locus on the θ-axis and the radius r-axis, a result of polar coordinate conversion when the variation of the radius r in each discontinuous circle read from the image data is large, and (b) is (a) It is a graph which shows what connected the locus | trajectory of). (A) is a graph showing, as a locus on the θ-axis and the radius r-axis, a result of polar coordinate conversion when the variation of the radius r in each discontinuous circle read from the image data is large, and (b) is (a) It is a graph which shows what connected the locus | trajectory of). FIG. 24 is a block diagram illustrating an overall configuration of an image processing system, illustrating still another embodiment of the image processing method and the image processing apparatus according to the present invention. It is a figure which shows the image data of the Fresnel lens used in the said image processing system. It is a flowchart which shows the method of detecting the circular center position of the concentric circle in the said Fresnel lens in a rough search level. It is a figure which shows the circle center position when the circle center position of the concentric circle in the said Fresnel lens is detected by the rough search level. It is a flowchart which shows the method of detecting the circular center position of the concentric circle in the said Fresnel lens in a fine search level. (A) is a figure which shows the circular center position of the middle course when carrying out polar coordinate conversion by making the coordinate away from the true circle center into the origin in the case of detecting the circular center position of the concentric circle in the said Fresnel lens at a fine search level. (B) is a figure which shows the circular center position in the middle of the time when polar coordinate conversion was carried out by making the coordinate close | similar to a true circle center into an origin. (A)-(d) is a figure which respectively shows the state of the circle in the image data processed with an image processing apparatus. (A)-(d) is a figure which respectively shows the state of the circle in the image data processed with the conventional image processing apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1a Image processing system 1b Image processing system 2 Imaging device 3 Object 4 Fresnel lens 5 Drive apparatus (drive means)
DESCRIPTION OF SYMBOLS 10 Image processing system main body 11 Control apparatus 12 Recording apparatus 13 Output apparatus 20 Image processing apparatus 21 Circle center temporary setting part (circle center temporary setting means)
22 Polar coordinate converter (polar coordinate converter)
23 One-dimensional projection unit (one-dimensional projection means)
24 evaluation value calculation unit (evaluation value calculation means)
25 circle center determination unit (circle center determination means)
30 Image processing device 31 Smoothing processing unit (smoothing processing means)
32 Edge processing unit (edge processing means)
33 Expansion / contraction processing unit (expansion / contraction processing means)
D Origin candidate area TH threshold

Claims (17)

An image processing apparatus for obtaining a center position of a circle displayed by image data,
A circular center temporary setting means for performing a Hough transform on the image data and temporarily setting the center position of the detected circle;
A polar coordinate conversion means for setting a predetermined range centered on the coordinates of the center position of the temporarily set circle, and converting the detected circle into a polar coordinate with each coordinate existing in the predetermined range as an origin;
One-dimensional projection means for one-dimensionally projecting the radius r of the circle obtained when the detected circle is converted into a polar coordinate with each coordinate as the origin in the direction of the central angle θ axis;
From the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection by the one-dimensional projection means, an evaluation value calculation for obtaining an evaluation value indicating linearity of the radius r of the circle with respect to the central angle θ-axis direction is calculated. Means,
Among the evaluation values from the distribution of the radius r of the circle obtained by performing polar coordinate conversion on the detected circle with the coordinates existing in the predetermined range as the origin, the coordinates of the origin where the evaluation value is highest are the coordinates of the circle. An image processing apparatus comprising: a circle center determining unit that determines a center position.   The image processing apparatus according to claim 1, further comprising a smoothing processing unit configured to perform a smoothing process on the image data before performing a Hough transform.   Edge processing means for detecting an edge in the radius r-axis direction and performing binarization processing with a predetermined value to extract an edge before one-dimensional projection on the circle obtained by the polar coordinate conversion is provided. The image processing apparatus according to claim 1, wherein the image processing apparatus is an image processing apparatus.   The evaluation value calculation means obtains the evaluation value by binarization processing using a threshold value from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by one-dimensional projection by the one-dimensional projection means. The image processing apparatus according to claim 1, 2, or 3.   The evaluation value calculation means obtains the evaluation value from a distribution in the central angle θ-axis direction at a radius r of a circle obtained by one-dimensional projection by the one-dimensional projection means using a threshold value by statistical data processing. The image processing apparatus according to claim 1, wherein the image processing apparatus is characterized.   The image processing apparatus according to claim 4, wherein the evaluation value calculating unit uses an inverse of a sum of luminance values of image data equal to or less than the threshold as an evaluation value indicating the linearity.   Prior to obtaining an evaluation value for a circle when polar coordinates are converted from a circle detected with each coordinate existing in the predetermined range as an origin, before performing an expansion process to connect the disconnected circles, and The image processing apparatus according to claim 1, further comprising an expansion / contraction processing unit that performs contraction processing to delete noise from a circle when polar coordinates are converted. An image processing method for obtaining a center position of a circle displayed by image data,
A circle center temporary setting step for performing a Hough transform on the image data and temporarily setting the center position of the detected circle;
A polar coordinate conversion step of setting a predetermined range centered on the coordinates of the center position of the temporarily set circle, and converting the detected circle into a polar coordinate with each coordinate existing in the predetermined range as an origin;
A one-dimensional projection step of one-dimensionally projecting the radius r of the circle obtained when the detected circle is converted into a polar coordinate with the respective coordinates as the origin in the direction of the central angle θ axis;
From the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step, evaluation value calculation for obtaining an evaluation value indicating linearity of the radius r of the circle with respect to the central angle θ-axis direction is calculated. Process,
Among the evaluation values from the distribution of the radius r of the circle obtained by performing polar coordinate conversion on the detected circle with the coordinates existing in the predetermined range as the origin, the coordinates of the origin where the evaluation value is highest are the coordinates of the circle. And a circle center determining step for determining the center position.   9. The image processing method according to claim 8, further comprising a smoothing process step of performing a smoothing process on the image data before performing a Hough transform.   An edge processing step of performing edge detection in the direction of the radius r-axis and extracting the edge by binarizing with a predetermined value before performing one-dimensional projection on the circle obtained by the polar coordinate conversion. 10. The image processing method according to claim 8, wherein the image processing method is characterized.   In the evaluation value calculation step, the evaluation value is obtained by binarization processing using a threshold value from the distribution in the central angle θ-axis direction at the radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection. The image processing method according to claim 8, 9 or 10.   In the evaluation value calculation step, the evaluation value is obtained from a distribution in the central angle θ-axis direction at a radius r of the circle obtained by the one-dimensional projection in the one-dimensional projection step, using a threshold value by statistical data processing. The image processing method according to claim 8, wherein:   The image processing method according to claim 11 or 12, wherein, in the evaluation value calculation step, an inverse number in a sum of luminance values of image data equal to or less than the threshold value is used as an evaluation value indicating the linearity.   Prior to obtaining an evaluation value for a circle when polar coordinates are converted from a circle detected with each coordinate existing in the predetermined range as an origin, before performing an expansion process to connect the disconnected circles, and 14. The image processing method according to claim 8, further comprising an expansion / contraction process step of performing a contraction process to delete noise from a circle when polar coordinates are converted. An image processing program for causing a computer to function as the image processing apparatus according to any one of claims 1 to 7,
An image processing program for causing a computer to function as each of the above means.   A computer-readable recording medium on which the image processing program according to claim 15 is recorded. In an image processing system that generates image data by imaging an object with an imaging device, and performs image processing on the image data,
Between the object and the imaging device, a Fresnel lens consisting of a convex lens of multiple concentric circles is provided,
The image processing apparatus according to claim 1, wherein the position of the optical axis of the Fresnel lens is calculated so that the optical axis of the Fresnel lens is disposed on an imaging target position of the object. ,
Based on the calculation result of the optical axis position of the Fresnel lens by the image processing device, driving means for moving the Fresnel lens so that the optical axis of the Fresnel lens is disposed on the imaging target position of the object. An image processing system characterized by being provided.

Images