JP5087151B2 - Image processing method, image processing apparatus, and image processing program - Google Patents

Description

  The present invention relates to an image processing method, an image processing apparatus, and an image processing program for extracting continuous line segments from a grayscale image, and in particular, extracts a linearly connected region from a grayscale image in consideration of the growth direction of the line segment. The present invention relates to an image processing method, an image processing apparatus, and an image processing program.

  With the recent demand for advanced personal authentication technology, many technologies for personal authentication using image data obtained by imaging a living body (subject) have been provided. For example, a part such as a fingerprint of a limb, a retina of the eye, a face, or a blood vessel that can distinguish an individual is imaged, and these characteristic parts are extracted from the captured image, and personal authentication is performed. The part suitable for such personal authentication is most suitable a part composed of relatively continuous line segments.

  On the other hand, since the captured image has a relatively low contrast and includes noise depending on the surrounding environment and imaging state, a technique is required for the technique for accurately extracting such continuous line segments. As a technique for extracting continuous line segments from this image, it is effective to perform edge enhancement processing and morphology processing for tracking line segments.

  Conventionally, it has been proposed to extract a line segment using a morphological function or a Gaussian Laplacian filter after binarizing a captured image using this morphological technique (see Patent Document 1). However, applying to an image after binarization does not use grayscale data obtained from a captured image, so that accurate line segment detection is difficult.

  Conventionally, as a method of performing morphological processing on grayscale data, grayscale data is subjected to open processing and top-hat processing, which is one of morphological processing, in order to extract line segments such as blood vessel images from the retina image of the eye. Has been proposed (see Non-Patent Document 1).

Japanese Patent Laying-Open No. 2004-329825 (FIG. 3)

Paper “Segmentation of Vessel-Like Pattern Using Mathematical Morphology and Curvature Evaluation” (F. Zana, JC Klein, IEEE Trans. Image Processing, vol. 10, pp1010-1019, July 2001)

  However, the morphological process is effective for extracting connected line segments, but if the contrast of the image is low, line segments that are not connected are also extracted, resulting in a decrease in line segment extraction accuracy.

  Accordingly, an object of the present invention is to provide an image processing method, an image processing apparatus, and an image processing program for extracting connected line segments using morphological processing from grayscale data of an image even if there is a difference in contrast between images. Is to provide.

  Another object of the present invention is to provide an image processing method, an image processing apparatus, and an image processing program for extracting connected line segments using morphological processing from grayscale data of an image without being affected by an imaging environment. It is to provide.

  Furthermore, another object of the present invention is to provide an image processing method, an image processing apparatus, and an image processing program suitable for extracting connected line segments from a gray scale data of a living body captured image using morphological processing. It is in.

In order to achieve this object, the image processing method for extracting a line segment element from a grayscale captured image according to the present invention scans an image according to the captured image in a plurality of directions, and extracts the extracted image from each direction. Performing a morphological process of extracting linear elements; a relatively high contrast ratio area above a relatively high first brightness level of the morphologically processed image ; and a relatively low second brightness level or above. Extracting a region having a relatively low contrast ratio; and extracting a linear element in the region having a relatively low contrast ratio where the region having a relatively high contrast ratio exists as a continuous line segment element.

An image processing apparatus for extracting line segment elements from a grayscale captured image according to the present invention includes: an imaging apparatus that captures a subject; and an operator in a plurality of directions with respect to an image according to the captured image captured by the imaging apparatus. And performing a morphological process for extracting linear elements from the extracted image in each direction, and a relatively high contrast ratio region of a relatively high first luminance level of the morphologically processed image ; A region having a relatively low contrast ratio that is equal to or higher than the low second luminance level is extracted, and the linear element in the region having the relatively low contrast ratio in which the region having the relatively high contrast ratio exists is used as a continuous line segment element. It has a line segment extraction device for extraction.

  Furthermore, the present invention preferably further includes a step of binarizing the image according to the captured image and selecting an extraction region of the captured image on which the morphological processing is executed.

  Furthermore, in the present invention, preferably, the method further includes a step of smoothing and edge enhancement processing the grayscale captured image to create an image according to the captured image.

  Furthermore, in the present invention, preferably, the extracted line segment element is smoothed and edge-enhanced, and the smoothed and edge-enhanced line segment element is binarized to generate line segment data. The method further includes the step of:

  In the present invention, it is preferable that the step of executing the morphological processing includes the step of scanning the operator in a plurality of directions to create an open processed image in each direction, and an image in each direction from the image according to the captured image. The method includes a step of creating a top hat processed image in each direction from an open processed image, and a step of extracting the linear elements by adding the top hat processed image in each direction.

  In the present invention, it is preferable that the step of extracting the line segment element includes a step of designating a region having a high contrast potential of the image according to the captured image, and the designation of the morphologically processed image. Extracting a region having a relatively high contrast ratio and a region having a relatively low contrast ratio with respect to the region, and a linear element of the region having a relatively low contrast ratio connected to the region having a relatively high contrast ratio, Extracting as line segment elements.

  In the present invention, it is preferable that the step of extracting the region includes a step of calculating a frequency histogram of luminance of the designated region, a region having a relatively high contrast ratio from the frequency histogram, and a relatively contrast ratio. And extracting a low region.

  Furthermore, in the present invention, it is preferable that the extraction step includes a relatively low brightness level as a relatively low contrast level area as a relatively low contrast ratio area and a relatively high contrast ratio area from the frequency histogram. Extracting a high marker area.

  Furthermore, in the present invention, it is preferable that the step of extracting as the line segment element includes a step of extracting a mask area having the marker area as the line segment element.

In the present invention, a region having a relatively high contrast ratio of a relatively high first luminance level and a region having a relatively low contrast ratio of a relatively low second luminance level are extracted from the morphologically processed image. Since the line elements in the relatively low contrast ratio area where there is a relatively high contrast ratio area are extracted as continuous line segment elements , the contrast ratio is continuously low from the high contrast ratio portion in the line segment growth direction. Since the portion is also extracted as one line segment, the line segment can be accurately extracted regardless of the contrast ratio.

1 is a configuration diagram of an image processing apparatus according to an embodiment of the present invention. It is a block diagram of the line segment extraction part of FIG. It is explanatory drawing of the captured image example of FIG. 1, FIG. It is explanatory drawing of the example of an image after the LoG Filter process of FIG. It is explanatory drawing of the area | region selection image example from the image of FIG. It is explanatory drawing of the example of an image after the morphological process of FIG. It is explanatory drawing of the example of an image after the extraction object selection process of FIG. It is explanatory drawing of the example of an image after the LoG Filter process of FIG. It is explanatory drawing of the example of an image after the binarization process of FIG. It is explanatory drawing of the LoG Filter process of FIG. FIG. 3 is a flowchart of region selection processing and top hat summation processing in FIG. 2. It is explanatory drawing of the area | region selection process of FIG. It is explanatory drawing of the opening process of FIG. It is explanatory drawing of the area | region selection process of FIG. 11, and a top hat summation process. FIG. 12 is an explanatory diagram of the top hat summation process of FIG. 11. FIG. 3 is an extraction area selection processing flowchart of FIG. 2. It is explanatory drawing of the area | region designation | designated process of FIG. It is explanatory drawing of the histogram creation of FIG. 16, a mask area | region, and a marker area | region detection process. It is explanatory drawing of the reconstruction process of FIG. It is explanatory drawing of the LoG Filter process of FIG. It is explanatory drawing of the line segment extraction operation | movement by the LoG Filter process of FIG.

  Hereinafter, embodiments of the present invention will be described in the order of an image processing apparatus, an image processing method, and other embodiments.

** Image processing device **
1 is a configuration diagram of an image processing apparatus according to an embodiment of the present invention, FIG. 2 is a configuration diagram of a line segment extraction unit in FIG. 1, and FIGS. 3 to 9 are images of a line segment extraction unit in FIG. It is explanatory drawing of a process example. 3 to 9 show an example of line segment extraction processing of a skin surface pattern as an object image, but it can also be applied to extraction of a line segment of a living body such as a retinal blood vessel image.

  As shown in FIG. 1, the imaging device 2 images the subject 5. The imaging apparatus 2 includes a light irradiation unit 3 that irradiates a subject 5 with light having a desired wavelength, an optical imaging unit 4 that receives reflected light from the subject 5 and converts the light into an electrical signal (image signal). Have The optical imaging unit 4 includes, for example, an optical mechanism such as a lens and an image sensor (for example, a CMOS image sensor) that converts a received light image into an image signal.

  The image sensor of the optical system imaging unit 4 is composed of, for example, an image sensor having 640 pixels in the horizontal direction and 480 pixels in the vertical direction, and outputs an electric signal having a magnitude corresponding to the amount of light received by each pixel to the line segment extracting unit 1. The line segment extraction unit 1 converts an image signal (analog signal) from the image sensor of the optical system imaging unit 4 into a grayscale digital signal, and extracts a line segment from the converted digital image signal.

  The line segment extraction process of the line segment extraction unit 1 will be described with reference to FIG. The captured image is stored as a grayscale digital signal. The LOG Filter process 40 smooths grayscale image data, removes noise, and differentiates twice to enhance the edges of the image. For example, when the LOG Filter process 40 is executed on a grayscale captured image (for example, a captured image of the skin surface) G1 as shown in FIG. 3, noise is removed by smoothing (integration), and the image is differentiated by differentiation. The edge is emphasized, and an image (blurred line segment image) G2 as shown in FIG. 4 is obtained.

  In the binarization region selection processing 42, the image G2 is binarized with a predetermined threshold value, and a region to be subjected to top hat summation processing which is one of morphological processing described later is selected. For example, when the image G2 subjected to the LOG Filter process as shown in FIG. 4 is binarized, an area selection image as shown in FIG. 5 is obtained. In FIG. 5, the black portion is selected as an area for performing the top hat summation process 44.

  Next, the top hat summation process 44 extracts continuous linear components from the image G2 subjected to the LOG Filter process. Therefore, the top hat summation process 44 scans a predetermined length of pixels (referred to as an element or an operator) in a predetermined direction with respect to the image G2, and creates an upper envelope image that allows the operator to enter in a direction with a high luminance level. An opening process 44-1 to perform, a top hat image process 44-2 for subtracting the upper envelope image from the original image G2, creating a top hat image, and adding a plurality of top hat images in the scanning direction in units of pixels; Consists of.

  For example, when the top hat summation process is performed on the image G2 in FIG. 4, as shown in FIG. 6, the isolated line segment is removed, and an image G4 having only a continuous linear component (white portion in FIG. 6) is obtained. . Here, in the present embodiment, the top hat summation process 44 is performed only on the above-described area selected portion (FIG. 5), that is, on an area where a continuous line segment may exist.

  If the top hat summation process 44, which is a morphological process, is performed in a plurality of directions on the entire image (300,000 pixels in the above example), the processing time becomes long. Since the number of line segments is an unspecified number, it is preferable that the number of scanning directions is larger in order to extract continuous line segments with high accuracy. For example, scanning in 12 directions (every 30 degrees) is preferable. In this case, the time for the continuous line segment extraction process becomes longer. In order to shorten the extraction processing time, it is effective to perform the top hat summation processing 44 on an area in which a continuous line segment may exist as in the present embodiment.

  Next, when grayscale data is targeted, there may be a portion having a different contrast ratio in a continuous line segment. For this reason, when binarizing the line segment image subjected to the top hat summation processing and extracting the line segment, the low contrast ratio portion is continuous with the high contrast ratio portion. Not extracted as part of a line segment.

  In this embodiment, the extraction target region selection process 46 is performed in order to extract a continuous line segment having such a different contrast ratio. The extraction target region selection process 46 is an extraction process in which a portion having a low contrast ratio that is continuous in a growth direction of a line segment from a portion having a high contrast ratio is also taken as one line segment.

  For this reason, the extraction target region selection processing 46 includes an extraction range selection processing 46-1 for selecting a region with a relatively high contrast ratio of the image G 4 subjected to the top hat summation processing 44 as an extraction range, and an extraction range selection processing 46. A histogram creation process 46-2 for creating a histogram of the extraction range selected in -1, and a region having a relatively high luminance level is selected from the histogram as a mask region, and a region having a higher luminance level is selected in the mask region. A mask area / marker area selection process 46-3 to be selected as a marker area, and a reconstruction process 46- to reconstruct a line segment continuous in the growth direction from the marker area within the selected mask area from the line segment of the image G4. It consists of four.

  For example, when the line segment of the image G4 in FIG. 6 is subjected to extraction target selection processing, as shown in the image G5 in FIG. It is reconstructed as a minute (white part in FIG. 7). Here, in this embodiment, by performing this processing 46, continuous line segments having the same contrast ratio can be obtained.

  The reconstructed image G5 is subjected again to LoG filter processing 47, smoothed and edge-enhanced, and the line segment image G6 subjected to smoothing and edge-enhanced in FIG. 8 is obtained. This line segment image G6 is binarized by a binarization process 48 with a predetermined threshold value to obtain a binary line segment image G9 shown in FIG. Then, a line segment having a certain width is thinned by a thinning process 49 to obtain a line segment image having a predetermined width (for example, one pixel).

  As described above, when extracting a continuous line segment by morphological processing, the range of the morphological process is limited to an area where the continuous line segment of the image may exist, thereby processing the morphological process which takes time. You can save time.

  In addition, when extracting a line segment from the obtained line segment image by morphological processing, if a region having a low contrast ratio is also continuous with a region having a high contrast ratio, it is reconstructed as a continuous line segment. Therefore, it is possible to accurately extract continuous line segments regardless of the contrast ratio.

** Image processing method **
Next, the line segment extraction process described in FIG. 2 will be described. FIG. 10 is an explanatory diagram of the LoG Filter process 40 of FIG.

  Hereinafter, the LoG Filter process 40 will be described assuming that the input image is f and the image brightness on the xy coordinates (image sensor surface) is f (x, y). The two-dimensional Gaussian function G (x, y) is defined by the following formula (1).

  By convolution of the Gaussian function G and the input image f, a smoothed image F (x, y) is obtained by the following equation (2).

  The smoothed image F (x, y) is partially differentiated twice, and the output g (x, y) of the LoG Filter is obtained by the following equation (3).

  In Expression (3), ∇ (Nabula) indicates partial differentiation, and in Expression (3), it indicates that the smoothed image F (x, y) is differentiated twice. That is, as is well known, the LoG Filter process 40 smoothes an image by integration, and emphasizes an edge by twice differentiation.

  The operation of the LoG Filter process 40 will be described with reference to FIG. FIG. 10 is described with a model diagram for the sake of simplicity. The three-dimensional coordinate x-axis and y-axis are two-dimensional plane coordinates, and the s-axis is a luminance level. When the LoG Filter process 40 is performed on the input image A1 (f), a three-dimensional image A2 (g) that is smoothed and edge-enhanced is obtained. Examples of this image are shown in FIGS.

  Next, the binarization region selection processing 42 and the top hat summation processing 44 will be described. FIG. 11 is a process flow diagram of the binarization region selection processing 42 and the top hat summation processing 44, FIG. 12 is an explanatory diagram of the binarization region selection processing operation of FIG. 11, and FIG. FIG. 14 is an explanatory diagram of the opening process of the mention process, FIG. 14 is an explanatory diagram of the top hat summation process, and FIG. 15 is an explanatory diagram of the top hat process.

  The process of FIG. 11 will be described with reference to FIGS.

  (S10) As described above, the LoG Filter process 40 is executed on the input image to obtain the LoG Filter image.

  (S12) Next, this image is binarized with a predetermined luminance slice value Th, and an opening processing target region is selected. In FIG. 12, two-dimensional coordinates (x or y and s) are shown for ease of explanation. The input image is sliced with a predetermined luminance slice value Th, and an area equal to or larger than the slice value is determined as the opening processing range. In other words, when the luminance level of the target line segment is high, there is almost no possibility that the target line segment exists in the low luminance level region. Conversely, the target line segment exists in the high luminance level region. There is a high possibility of doing. For this reason, the time required for the hat top simplification processing which requires a long processing time is shortened by setting the processing range of the hat top summation processing 46 for extracting the line segment not only the entire image but only the region where the luminance level of the image is high.

  (S14) The image after the LoG Filter processing is subjected to the opening processing in the designated opening area (black portion in FIG. 5). Although the opening process is well known, it will be briefly described with reference to the model diagrams of FIGS.

  As shown in FIG. 13, the operator (or element) B is scanned with respect to the input image A3 (f (x)). The operator B has a predetermined length. FIG. 13 shows an example in which the operator B is scanned in three directions. As shown in FIG. 15, when the operator B is scanned in the x-axis direction with respect to the input image f (x), the input image f (x) is divided into a portion where the operator B can enter and a portion where the operator B cannot enter. The process of obtaining the envelope γB (f) (x) of the portion where the operator B can enter is called a linear opening process.

  (S16) Using the linearly-opened image, top hat processing is performed on the input image. The top hat process is performed by the calculation of the following equation (4).

f (x) −γB (f) (x) (4)
That is, a function of a portion (this is called a top hat) where operator B in FIG. 15 cannot enter is obtained.

  (S18) The image after the top hat process obtained by scanning the operator in each direction is added in units of pixels.

  The opening process and the top hat summation process will be described with reference to the image example of FIG. For simplification of description, FIG. 14 will be described with an example of extracting linear components extending in the x and y directions for image data obtained by binarizing the image B1. In the image B1, the hatched portion is a region with a high luminance level. When this image is binarized, a binarized input image is obtained. Here, the horizontal axis is x and the vertical axis is y. Further, the operator Bx in the x direction will be described as an operator having a length of 3 pixels in the x direction, and the operator By in the y direction will be described as an operator having a length of 3 pixels in the y direction.

  First, the operator Bx is scanned in the x direction of the input image, and when the trajectory that the operator Bx can enter as described above is obtained, an x-axis linear opening image is obtained. That is, a pixel value “1” is assigned to a portion where three pixels “1” are continuous in the x-axis direction of the input image, and the others are pixel values “0”.

  Using this x-axis linear opening image and the input image, an x-axis top hat image is obtained for each pixel by equation (4). In contrast to the original image B1, this image shows the contour of a continuous linear component in the x-axis direction.

  Similarly, first, when the operator By is scanned in the y direction of the input image, and the locus by which the operator By can enter is obtained as described above, a y-axis linear opening image is obtained. That is, a pixel value “1” is assigned to a portion where three pixels “1” are continuous in the y-axis direction of the input image, and the others are pixel values “0”.

  Using this y-axis linear opening image and the input image, a y-axis top hat image is obtained for each pixel according to Equation (4). In contrast to the original image B1, this image shows a continuous linear component in the y-axis direction.

  The x-axis top hat image and the y-axis top hat image are added in units of pixels to obtain a top hat summation image. In contrast to the original image B1, the isolated linear component indicated by the black circle in the image B1 is removed, and the continuous linear component of the original image B1 is extracted.

  Next, the extraction target area selection processing 46 will be described with reference to FIGS. 16 is an extraction target area selection processing flowchart, FIG. 17 is an explanatory view of the extraction range selection process of FIG. 16, FIG. 18 is an explanatory view of the histogram creation process and mask / marker area selection process of FIG. These are explanatory drawings of the reconstruction process of FIG.

  Hereinafter, the extraction target area selection process of FIG. 16 will be described with reference to FIGS. 17 to 19.

  (S20) The region having a relatively high contrast ratio of the image G4 subjected to the top hat summation process 44 is selected as the extraction range X. As shown in FIG. 17, the central portion tends to have a higher contrast ratio than the peripheral portion of the image Y. This is related to the imaging conditions. That is, depending on the performance of the optical system, the light irradiation state of the subject, and the external light condition of the imaging apparatus, the peripheral portion tends to be blurred. Therefore, the extraction range selection process 46-1 selects the region X including the central portion excluding the peripheral portion of the image Y as the target region as shown in FIG.

  (S22) Next, a histogram of the extraction range X selected in the extraction range selection process 46-1 is calculated. As shown in FIG. 18, the horizontal axis represents the luminance level, the vertical axis represents the pixel frequency, and the number of pixels at each luminance level is calculated as a luminance histogram.

  (S24) As shown in FIG. 18, from this luminance histogram distribution, the luminance level M1 at which the frequency number is about half of the total frequency number is set as the threshold value of the mask area, and the luminance level M1 or higher on the histogram Are selected as a mask region. Next, the luminance level M2 at which the frequency number is about half of the overall frequency number is set as the threshold value of the marker region, and pixels having the luminance level M2 or higher on the histogram are selected as the marker region. That is, an area having a relatively high luminance level is selected as a mask area from the histogram, and an area having a higher luminance level is selected as a marker area within the mask area.

  (S26) Using this mask area and marker area, an image subjected to top hat summation processing is reconstructed. That is, as shown in FIG. 19, when the selected mask area and the pixels of the marker area are imaged, an image B3 having mask areas C1 and C3 and a marker area C2 is obtained. As shown in FIG. 19, the reconstruction is performed with the mask region C1 in which the marker region C2 exists as a line segment D1. Here, the line segment C3 having no marker region is removed.

  As described above, even when the contrast ratio (the level difference between the line segment portion and the other portion) is low, it is extracted as one line segment when it is connected to a line segment element with a high contrast ratio. If such processing 46 is not executed, the extracted line segments will be C1 and C3 or only C2, as will be described with reference to FIG. 19, and the extracted line segments will differ depending on the contrast ratio.

  In addition, since the marker region and the mask region are determined at the pixel level of the region X with a high contrast ratio, the marker region and the mask region can be determined by removing the influence of noise in the peripheral portion with a low contrast ratio.

  Further, since the marker region and the mask region are determined by the frequency histogram of the luminance level of the pixel of the image, even if the line segment pattern of the image is different, the marker region and the mask corresponding to the image, that is, the relative marker region and mask. You can select an area.

  Next, the LoG filter process 47 will be described with reference to FIGS. The LoG Filter process 47 is the same as the contents of the LoG Filter process 40, and integrates, differentiates and edge-enhances. However, the LoG filter processing 47 in the latter stage after the line segment extraction aims to smooth the edge of the line segment D1 obtained as shown in FIG. That is, as shown in FIG. 20, the original image after integration is differentiated once and twice (differentiated twice). As a result, as shown in FIG. 21, a jagged portion D3 around the line segment D1 is smoothed, and a smooth line segment element D2 can be extracted.

  This smooth line segment element is binarized 48 and further thinned 49 to obtain a skeletonized line segment. The LoG Filter process 47 described above is effective in binarizing and thinning, and can be easily thinned.

** Other embodiments **
In the above-described embodiment, the operator of the top hat summation processing 44 has been described as having a width of 1 pixel and a length of 3 pixels. Depending on the accuracy, it can be selected as appropriate. For example, it may be about 20 pixels. Further, the shape of the operator is not limited to a rectangle, and other shapes such as an ellipse can be applied.

  The number of scanning directions of the operator can be variously selected according to the direction, accuracy, and processing time of the segment to be extracted, and is preferably four or more. In addition, if the scanning direction is small and the processing time is sufficient, the region selection processing 42 by binarization may be omitted. Similarly, when the same contrast ratio is targeted, the extraction target selecting process 46 can be omitted.

  Furthermore, the top hat summation process described as the morphological process can be performed by other methods as long as the required elements or operators are scanned in a plurality of directions with respect to the image and linear elements are extracted from the extracted images in the respective directions. Applicable. Furthermore, the application field is not limited to the surface of the skin, but can be applied to a blood vessel image of a living body, a pattern of a living body, and the like.

  As mentioned above, although this invention was demonstrated by embodiment, in the range of the meaning of this invention, this invention can be variously deformed, These are not excluded from the scope of the present invention.

  (Supplementary note 1) In an image processing method for extracting line segment elements from a grayscale captured image, an operator scans an image according to the captured image in a plurality of directions, and extracts linear elements from the extracted image in each direction. Performing a morphological process; extracting a region having a relatively high contrast ratio and a region having a relatively low contrast ratio from the morphologically processed image; and connecting the relatively contrast region to the region having a relatively high contrast ratio An image processing method comprising: extracting a linear element in a low ratio area as a line segment element.

  (Supplementary note 2) The image processing method according to supplementary note 1, further comprising a step of binarizing an image according to the captured image and selecting an extraction region of the captured image on which the morphological processing is executed.

  (Supplementary note 3) The image processing method according to supplementary note 1, further comprising the step of smoothing and edge enhancement processing the captured image of the gray scale to create an image according to the captured image.

  (Supplementary Note 4) The method further includes the steps of smoothing the extracted line segment element and performing edge enhancement, and binarizing the smoothed and edge enhanced line segment element to generate line segment data. The image processing method according to Supplementary Note 1, wherein

  (Supplementary Note 5) The step of executing the morphological process includes: scanning the operator in a plurality of directions to create an open process image in each direction; and an open process image in each direction from an image according to the captured image. The image processing method according to appendix 1, characterized by comprising a step of creating a top hat processed image in each direction and a step of extracting the linear elements by adding the top hat processed images in each direction.

  (Supplementary note 6) The step of extracting the line segment element is compared with the step of designating a region having a high contrast potential of the image according to the captured image and the designated region of the morphologically processed image. Extracting a region having a high contrast ratio and a region having a relatively low contrast ratio, and extracting linear elements of the region having a relatively low contrast ratio connected to the region having a relatively high contrast ratio as line segment elements The image processing method according to claim 1, further comprising the step of:

  (Supplementary note 7) The step of extracting the region includes a step of calculating a luminance frequency histogram of the designated region, and extracting a region having a relatively high contrast ratio and a region having a relatively low contrast ratio from the frequency histogram. The image processing method according to claim 6, further comprising the step of:

  (Supplementary Note 8) The extraction step includes, as a region having a relatively low contrast ratio from the frequency histogram, a mask region having a relatively low brightness level and a marker region having a relatively high brightness level as a region having a relatively high contrast ratio. The image processing method according to claim 7, further comprising a step of extracting.

  (Supplementary note 9) The image processing method according to supplementary note 8, wherein the step of extracting the line segment element includes a step of extracting a mask area having the marker area as the line segment element.

  (Supplementary Note 10) In an image processing apparatus that extracts line segment elements from a grayscale captured image, an operator is moved in a plurality of directions with respect to an imaging apparatus that captures a subject and an image that follows the captured image captured by the imaging apparatus. Scan and execute morphological processing to extract linear elements from the extracted images in each direction, extract regions with relatively high contrast ratio and regions with relatively low contrast ratio from the morphologically processed image, and compare An image processing apparatus comprising: a line segment extracting device that extracts a line element in a region having a relatively low contrast ratio connected to a region having a high contrast ratio as a line segment element.

  (Additional remark 11) The said line segment extraction apparatus binarizes the image according to the said captured image, and selects the extraction area | region of the said captured image which performs the said morphology process, The image processing apparatus of Additional remark 10 characterized by the above-mentioned.

  (Supplementary note 12) The image processing device according to supplementary note 10, wherein the line segment extraction device smoothes and edge-enhances the grayscale captured image to create an image according to the captured image.

  (Supplementary Note 13) The line segment extraction device smoothes the extracted line segment element and performs edge enhancement processing, and binarizes the smoothed and edge-enhanced line segment element to create line segment data The image processing apparatus according to appendix 10, wherein:

  (Supplementary note 14) The line segment extraction device scans the operator in a plurality of directions, creates an open process image in each direction, and creates an open process image in each direction from an image according to the captured image. The image processing apparatus according to appendix 10, wherein a top hat processed image is created, and the top hat processed image in each direction is added to perform a morphological process for extracting the linear element.

  (Supplementary note 15) The line segment extraction device includes a region having a relatively high contrast ratio and a region having a relatively low contrast ratio with respect to a region having a high contrast ratio of the captured image of the morphologically processed image. The image processing apparatus according to appendix 10, wherein a linear element in the region having a relatively low contrast ratio extracted and connected to the region having a relatively high contrast ratio is extracted as a line segment element.

  (Supplementary Note 16) The line segment extraction device calculates a frequency histogram of luminance of the designated region, and extracts the region having a relatively high contrast ratio and the region having a relatively low contrast ratio from the frequency histogram. The image processing apparatus according to supplementary note 15 as a feature.

  (Supplementary Note 17) The line segment extraction device includes a mask region having a relatively low brightness level as a region having a relatively low contrast ratio and a marker having a relatively high brightness level as a region having a relatively high contrast ratio from the frequency histogram. The image processing apparatus according to appendix 16, wherein an area is extracted.

  (Additional remark 18) The said line segment extraction apparatus extracts the mask area | region which has the said marker area | region as said line segment element, The image processing apparatus of Additional remark 17 characterized by the above-mentioned.

A region having a relatively high contrast ratio of a relatively high first luminance level or higher and a region having a relatively low contrast ratio of a relatively low second luminance level or higher are extracted from the morphologically processed image. In order to extract a linear element in a region having a relatively low contrast ratio and a region having a high contrast ratio as a continuous line segment element, a portion having a low contrast ratio that is continuous in the growth direction of the line segment from a portion having a high contrast ratio is also 1 Since it is extracted as one line segment, it can be extracted with high accuracy regardless of the contrast ratio. For this reason, continuous line segments can be extracted from a blurred image with high accuracy.

DESCRIPTION OF SYMBOLS 1 Line segment extraction part 2 Imaging device 3 Irradiation part 4 Imaging part 5 Subject 40,47 LoG Filter process 42 Area designation | designated process by binarization 44 Top hat siment process 46 Extraction object selection process 48 Binarization process 49 Thinning processing

Claims (7)

In an image processing method for extracting line segment elements from a grayscale captured image,
Performing an morphological process of scanning an operator according to the captured image in a plurality of directions and extracting a linear element from the extracted image in each direction;
Extracting a relatively high contrast ratio region above a relatively high first luminance level and a relatively low contrast region above a relatively low second luminance level of the morphologically processed image ;
An image processing method comprising: extracting a linear element in a region having a relatively low contrast ratio where the region having a relatively high contrast ratio exists as a continuous line segment element. Extracting the region having a relatively high contrast ratio and the region having a relatively low contrast ratio,
Calculating a frequency histogram representing the number of pixels at each luminance level of the morphologically processed image;
The image processing method according to claim 1 , further comprising: setting the second luminance level and the first luminance level from the frequency number of the frequency histogram . Further comprising the step of designating a region with a high contrast potential of the image according to the captured image;
The step of extracting the region having a relatively high contrast ratio and the region having a relatively low contrast ratio includes a step of extracting a region having a relatively high contrast ratio and a region having a relatively high contrast ratio with respect to the designated region of the morphologically processed image. the image processing method according to claim 1, characterized in that it comprises a step of extracting a lower region. In an image processing apparatus that extracts line segment elements from a grayscale captured image,
An imaging device for imaging a subject;
An operator scans an image according to the captured image captured by the imaging device in a plurality of directions, executes a morphological process for extracting a linear element from the extracted image in each direction, and relatively compares the morphologically processed image . A region having a relatively high contrast ratio above a high first luminance level and a region having a relatively low contrast ratio above a relatively low second luminance level are extracted, and the region having a relatively high contrast ratio exists . An image processing apparatus comprising: a line segment extraction device that extracts linear elements in a region having a relatively low contrast ratio as continuous line segment elements. The line segment extraction apparatus designates a region having a high contrast of an image according to the captured image, and compares the designated region of the morphologically processed image with a region having a relatively high contrast ratio. The image processing apparatus according to claim 4, wherein a region having a low contrast ratio is extracted. The line segment extraction device smoothes the extracted line segment element and performs edge emphasis processing, and binarizes the smoothed and edge emphasized line segment element to create line segment data. The image processing apparatus according to claim 4. An image processing program for extracting line segment elements from a grayscale captured image,
Performing an morphological process of scanning an operator according to the captured image in a plurality of directions and extracting a linear element from the extracted image in each direction;
Extracting a relatively high contrast ratio region above a relatively high first luminance level and a relatively low contrast region above a relatively low second luminance level of the morphologically processed image ;
An image processing program for causing a computer to execute a step of extracting linear elements in a region having a relatively low contrast ratio where the region having a relatively high contrast ratio exists as a continuous line segment element.