JP4834464B2 - Image processing method and image processing apparatus - Google Patents

Description

  The present invention relates to an image processing method and an image processing apparatus that extract a specific region in a digital image, and more particularly to an image processing method and an image processing device that extract a blood vessel region from an image of a living body.

  2. Description of the Related Art Conventionally, there is an apparatus that uses near infrared light as an apparatus for acquiring a blood vessel position of a living body as an image. This is because hemoglobin in blood utilizes the property of absorbing near-infrared light, irradiates the living body with light of near-infrared wavelength, and reflects or transmits light from the living body as near-infrared light wavelength. A black and white CCD (Charge Coupled Device) camera with high near-infrared sensitivity is imaged through an optical filter that passes through the region. At this time, since reflection or transmission is reduced only in the region where the blood vessel exists, the image appears dark on the image. Therefore, the blood vessel region is extracted by extracting the darkly projected region.

  As image processing for extracting a blood vessel region in a darkly projected region, a method of tracking pixels that are darkly projected is generally used. When the near-infrared light is not uniformly applied due to the curved surface of the living body and the brightness of the background changes greatly, the binarization processing and edge extraction processing based on the threshold simply extract the blood vessels with high accuracy. I can't. However, even if the brightness changes greatly, the blood vessel region is darker than the surroundings as long as it is limited to a local range. Therefore, the blood vessel region can be extracted by tracking the region darker than the surroundings (for example, patents). Reference 1). In the blood vessel extraction process used in the apparatus described in Patent Literature 1, in order to extract blood vessels whose number or length is unknown, the trajectory when moving from a certain position to a darker portion is changed to various positions. By obtaining many different lengths and overlaying them, the blood vessels are statistically raised.

  FIG. 17 is a flowchart showing blood vessel extraction processing (processing performed on a finger image obtained by imaging a finger) disclosed in Patent Document 1. In the figure, first, a score table having the same size as the image is prepared to hold the history of tracking blood vessels, and all of them are initialized to 0 (step S1700). In the blood vessel tracking loop (step S1701) repeatedly executed as many times as necessary to bring out the entire blood vessel pattern, the starting position of the tracking point for performing blood vessel tracking for one time is first determined by a random number ( Step S1702). However, since the blood vessel cannot be traced correctly if the background, the fingertip, the base of the finger, or the vicinity of the contour of the finger is used as a starting point, the contour information of the finger is used so that they do not become the starting point.

  It is easier to follow the blood vessel if the starting point is placed on the blood vessel. Therefore, a plurality of starting point candidates are determined, and the pixel having the darkest luminance value among them is set as the starting point. However, if the starting point is always determined under this condition, it is difficult to track the blood vessels in the bright part.Therefore, the pixel with the darkest luminance value among the multiple starting point candidates is not always set as the starting point, but with a certain probability. The pixel with the brightest luminance value is taken as the starting point. This probability is determined by a random number. Next, the direction in which the tracking point is easy to move is determined (step S1703). This property is used to determine the movable point described later. As an example of this determination method, it is determined by random numbers that it has a property of easily moving right or left and upward or downward. Subsequently, a “life” indicating the tracking length of the tracking point is determined (step S1704), and at the same time, the value is given to the tracking point as an initial value. The tracking is stopped when the tracking is performed for a distance determined by the lifetime. In other words, the tracking length is set as the lifetime of the tracking point, and each time one pixel is traced, the lifetime is reduced and the tracking is terminated when the lifetime is exhausted. This lifetime is determined using a random number.

  Subsequently, the tracking point is moved. First, the point at which the tracking point can move next is determined (step S1705). Many of the blood vessels run in the long axis direction of the finger, but the blood vessels are more emphasized by matching the tendency of the tracking point to move in the direction of the blood vessel running direction. Therefore, the tendency of movement of the moving point is controlled by giving a tendency to the candidate point that can be moved next.

  As an example of the movement tendency, in order to make it easy to move in the left and right major axis directions with a probability of 50%, the left or right three neighborhoods are set as movable points, and 30% of the remaining 50% are in the minor axis direction of the finger. In order to make it easier to move, the upper or lower three neighborhoods are set as movable points. Otherwise, the vicinity of 8 is set as a movable point. However, in any case, it cannot move to the outside of the trajectory or finger that has been followed. In this way, the movable point is obtained. If there is no movable point (step S1706), the tracking at the current tracking point is terminated.

  Subsequently, of the movable points, the pixel moves to the pixel having the darkest luminance value (step S1707). Then, the current position is registered and updated as trajectory information so that the trajectory that the current tracking point has traced in the past is not traced again (step S1708). At this time, the score is added to the position of the score table corresponding to the coordinates of the pixel (step S1710). Here, 5 points are added as an example. Further, the lifetime which is the tracking length of the tracking point is reduced by one (step S1711). Whether or not the tracking point has a lifetime is determined (step S1712). If there is a lifetime, the process returns to the determination of the movable point again (step S1705), and the movement, the addition of the score, and the update of the trajectory information are repeated. When the lifetime has expired, the trace information traced is initialized (step S1713), and the tracking at the current tracking point is completed. Such a blood vessel tracking process is repeated many times. When all of the repetitions are completed, the score of the score table corresponding to the position becomes higher as the pixel that has been traced more frequently, that is, the portion having a higher probability of being a blood vessel. Conversely, a position with a low score has a high probability that it is not a blood vessel. Therefore, the vein pattern itself appears in this score table. Therefore, by capturing this score table as an image, an image obtained by extracting only the vein pattern can be obtained.

  In order to make the vein pattern thus obtained easy to use for matching, each column of the score table as a vein pattern is classified according to the score. Here, it is classified into four types as an example (step S1714). First, it is assumed that there is no blood vessel at a pixel position with a low score. Further, it is assumed that a pixel position having a high score is likely to be a blood vessel. A pixel position having an intermediate score is an ambiguous region that may be a blood vessel but is not surely a blood vessel. Furthermore, the pixel located outside the outline of the finger is used as the background. By correlating these four types with luminance values, a vein pattern image can be obtained.

  Finally, in order to fill in the hole of the pixel that was not tracked by chance, the blood vessel portion and the ambiguous portion were subjected to expansion processing (step S1715). The dilation process examines the vicinity of 8 pixels of the blood vessel part or the ambiguous part for all pixels existing in the image, and if the number of pixels of the non-blood vessel part is 4 or less, the non-blood vessel part is ambiguous. This is done by converting into parts.

  Thus, in this blood vessel extraction process, a score table having the same size as the image is first prepared to hold a history of tracking blood vessels, and all of them are initialized to zero. As a process of performing blood vessel tracking for one time, first, the tracking point starting position, the direction in which the tracking point is easy to move, and the tracking length are set using random numbers as the initial value of the tracking point. At this time, the movable point where the tracking point can move next is determined in accordance with the tendency of the direction in which the blood vessel runs, and is moved to the pixel having the darkest luminance value among the determined movable points. At the same time, the score is added to the position of the score table corresponding to the coordinates of the pixel. These series of processes are repeatedly executed as many times as necessary to bring out the entire blood vessel pattern. In the score table created in this way, the higher the probability of being a blood vessel, that is, the higher the probability that it is a blood vessel, the higher the score corresponding to that position, so blood vessel extraction is performed by extracting the high score pixel. I was going.

JP 2002-83298 A

  However, in the configuration described in Patent Document 1 described above, when the blood vessel tracking is performed, the starting position of the tracking point, the tracking direction, and the tracking length are set with random numbers, so that blood vessel extraction can be performed stably. However, not all blood vessels in the image can be extracted depending on the random value. In addition, it is possible to extract blood vessels in a specific direction by matching the direction in which the tracking point is easy to move with the tendency of the direction in which the blood vessel runs. However, since the tracking direction is determined by a random number, it can be extracted reliably. Furthermore, there is a problem that the conventional method cannot be used to extract only blood vessels having specific shape characteristics such as length and thickness.

  The present invention has been made in view of such circumstances, and an object thereof is to provide an image processing method and an image processing apparatus capable of accurately extracting a blood vessel having a characteristic of an intended shape.

  In order to solve the above conventional problems, the image processing method of the present invention irradiates a living body with near-infrared light, images near-infrared wavelength components of reflected light or transmitted light, and digitizes the near-infrared light image. On the other hand, a dark region extraction step of extracting a pixel having a luminance value relatively smaller than that of surrounding pixels as a dark pixel and creating dark pixel image data in which a first feature amount is added to each pixel, and the dark region The dark pixels continuously distributed so as to match one or a plurality of preset mask patterns are extracted from the dark pixel image data created in the extraction step as continuous dark regions, and each pixel has a second feature. The second feature amount is obtained from a continuous dark region extraction step for creating continuous dark region image data in which a quantity is written, and a plurality of continuous dark regions in the continuous dark region image data created in the continuous dark region extraction step. Before the area larger than the predetermined value Extracted as a specific area where the blood vessel is present in the living body, and a specific area extracting step of creating a specific area image data.

  According to this method, a dark pixel having a luminance value smaller than that of a peripheral pixel in a near-infrared light image is extracted, and only a continuous dark pixel is extracted as a continuous dark region using a mask pattern therefrom. Since the blood vessel region is extracted based on the continuous dark region, the blood vessel region having the intended shape shown in the mask pattern can be accurately extracted with a small amount of calculation. Further, since the calculation amount does not increase or decrease depending on the shape of the blood vessel, stable blood vessel extraction can be performed with a constant processing load and processing time, and the present invention can be applied not only to still images but also to moving images.

  Furthermore, in the image processing method of the present invention, the dark region extracting step sequentially extracts a predetermined range of regions while shifting the position by one pixel relative to the near-infrared light image, and extracts at each position. Based on the luminance value of each pixel in the region, a dark pixel evaluation value for evaluating whether the central pixel located at the center of the region is a dark pixel is obtained as the first feature amount, and the obtained dark pixel evaluation The dark pixel image data is created based on the value.

  According to this method, it is possible to create dark pixel image data indicating dark pixels that are pixels having a luminance value smaller than that of the peripheral region in the near-infrared light image.

  Further, in the image processing method of the present invention, the dark region extraction step obtains a value obtained by subtracting the luminance value of the central pixel from the luminance value of all pixels in the extracted region for each pixel, and all the pixels after the subtraction Is the dark pixel evaluation value.

  According to this method, it is possible to reliably evaluate whether the central pixel is a dark pixel.

  Furthermore, in the image processing method of the present invention, in the dark region extraction step, a pixel in which the dark pixel evaluation value of the central pixel is larger than a predetermined threshold is set as the dark pixel, and the extracted region is When the center pixel is a dark pixel, the result obtained by dividing the dark pixel evaluation value of the dark pixel by a predetermined value is added as the feature amount of the dark pixel, and all the pixels other than the dark pixel have zero as the feature amount. In addition, when the center pixel is not a dark pixel, zero is added as a feature amount to all the pixels.

  According to this method, by dividing the dark pixel evaluation value of the dark pixel by a predetermined value, the value is reduced while maintaining the characteristic of the dark pixel evaluation value, and the image memory size of the dark pixel data to be secured in advance is reduced. It can be made small, and the calculation load can be reduced in the subsequent processing.

  Furthermore, in the image processing method of the present invention, the continuous dark region extracting step shifts the position where the mask pattern is superimposed with respect to the dark pixel image data by relatively shifting one pixel at a time. Continuous evaluation values for evaluating continuity are sequentially obtained, continuous dark regions are extracted based on the continuous evaluation values, and the continuous evaluation is performed on the corresponding dark pixels based on the position information of the continuous evaluation values and the mask pattern. A value is written as the second feature value to create continuous dark area image data.

  According to this method, a continuous dark region that is a dark pixel in which pixels are continuous is extracted from the dark pixel image data using a mask pattern that indicates pixel continuity, and a continuous dark region that indicates the continuous dark region is extracted. Create area image data. Since the dark pixel image data is obtained by extracting pixels that are darker than the surroundings as well as blood vessels, noise other than blood vessels is also included here. Therefore, by extracting dark pixels in which pixels are continuous, that is, dark pixels in a line shape that is a characteristic of blood vessels, it is possible to extract a region having a higher possibility of blood vessels.

  Furthermore, in the image processing method of the present invention, the continuous dark region extracting step assigns a different weight value to each mask pattern, and weights a continuous evaluation value obtained for each mask pattern.

  According to this method, the numerical value of the continuous evaluation value can be intentionally increased, and blood vessels in the longitudinal direction of the arm can be preferentially extracted.

  Furthermore, in the image processing method of the present invention, the continuous dark region extraction step includes, in the creation of the continuous dark region image data, a pixel position on the continuous dark region image data corresponding to a position matching the mask pattern, The result of dividing the continuous evaluation value by a predetermined value is written as the second feature amount. If the value has already been written, the result is written as the second feature amount only when the value to be written is a large value. I do.

  According to this method, by dividing the continuous evaluation value by a predetermined value, the value is reduced while maintaining the characteristics of the continuous evaluation value, and the image memory size of the continuous dark area image data secured in advance is reduced. And the calculation load can be reduced in the subsequent processing. In addition, when a value has already been written when writing a value to continuous dark area image data, the larger one is written to prevent overwriting with 0, and a continuous pixel sequence at that pixel position is prevented. It is possible to leave a continuous evaluation value indicating sex.

  Further, in the image processing method of the present invention, the continuous dark region extracting step includes the continuous dark region having the maximum continuous evaluation value among the continuous evaluation values and the mask pattern in the creation of the continuous dark region image data. And the continuous evaluation value is written as the second feature amount only to the corresponding dark pixel.

  According to this method, it is possible to extract only the dark pixels having the largest continuous evaluation value in the region where the mask patterns are overlapped, and it is possible to extract only the region particularly showing the blood vessel characteristics.

  Furthermore, in the image processing method of the present invention, the specific region extracting step uses, as the specific region, a pixel in which the second feature amount is larger than a predetermined threshold at each pixel position of the continuous dark region image data. Extraction is performed, and specific area image data is created based on the second feature amount.

  According to this method, a blood vessel region can be extracted by extracting a pixel having a large pixel value from continuous dark region image data. Furthermore, it is possible to know the position where a clearer blood vessel exists based on the size of the pixel value.

  Furthermore, in the image processing method of the present invention, the specific region extracting step is configured such that the number of pixels of the region extracted as the specific region from the continuous dark region image data is equal to or less than the predetermined number of pixels as the predetermined threshold. Calculate the correct value and set the value.

  According to this method, for example, the number of pixels in the region of the living body part to be imaged in advance and the ratio of the blood vessel region in the living body part are checked, and the number of pixels is calculated by multiplying the number of pixels in the living body part by the proportion of blood vessels. Then, a threshold value is set such that the number of pixels in the region extracted as the specific region is equal to or less than the calculated number of pixels. As a result, even if the brightness of the near-infrared light image picked up for reasons such as how the illumination light strikes, a blood vessel region having a certain number of pixels or more can be extracted constantly.

  Further, in the image processing method of the present invention, each of the specific region extraction step calculates a weight value according to an area value of a continuous dark region or a vertical / horizontal width ratio after creating the specific region image data. Pixels are weighted, and only pixel regions where the weighted values are larger than a predetermined threshold are extracted again as specific regions.

  According to this method, it is possible to remove noise such as body hair close to the characteristics of blood vessels, and the accuracy of blood vessel extraction can be further improved.

  Furthermore, in the image processing method of the present invention, the dark region extraction step further includes a calculation region calculation step of extracting a calculation region image excluding a background region other than the living body from the near-infrared light image before the dark region extraction step. In the extraction step and the continuous dark region extraction step, processing is performed based on the calculation region image.

  According to this method, by setting the calculation area, erroneous extraction of blood vessels in the background area can be eliminated, and the calculation amount can be reduced by narrowing the calculation range.

  Furthermore, in the image processing method of the present invention, the calculation area calculation step extracts a calculation area by comparing a luminance value of each pixel of the near-infrared light image with a predetermined threshold value, and generates the calculation area image. .

  According to this method, a calculation area excluding the background area can be obtained by extracting pixels having a luminance value larger than a predetermined threshold value.

  Furthermore, in the image processing method of the present invention, the calculation area calculation step sets the predetermined threshold value to a value such that the number of pixels extracted from the near-infrared light image is equal to or greater than a predetermined number of pixels.

  According to this method, for example, by setting a threshold value that separates the distribution of the background region and the living body region in advance, it is possible to obtain a calculation region that includes only the living body region excluding the background region.

  Furthermore, in the image processing method of the present invention, the calculation area calculation step compares the predetermined threshold value, and if there are a plurality of areas in which pixels that are candidates for the calculation area are gathered, an area having the maximum number of pixels Is the computation region image.

  According to this method, when there are a plurality of regions, a region having the maximum area value is set as a calculation region, whereby a calculation region composed of only a living body region can be obtained.

  Furthermore, in the image processing method of the present invention, the calculation area calculation step performs a correction to reduce the size of the calculation area with respect to the extracted calculation area.

  According to this method, for example, when imaging is performed by irradiating near-infrared light on the arm, near-infrared light is not sufficiently applied to the side surface of the arm, and blood vessel characteristics do not appear. Further, when the inside of the arm is imaged, since there is body hair near the side surface, it is better to remove the vicinity of the side surface from the calculation area. Therefore, the side pixels can be excluded from the calculation area by shrinking the calculation area inward by several pixels.

  The image processing method of the present invention further includes an image conversion step of correcting the near-infrared light image before the dark region extraction step, and the size of the living body region in the near-infrared light image is constant. Correction for enlarging or reducing the size of the near-infrared light image is performed so as to obtain a ratio.

  According to this method, since it can be converted into an image suitable for blood vessel extraction processing for extracting blood vessels, a blood vessel region can be extracted with stable accuracy.

  Furthermore, the image processing method of the present invention further includes an image conversion step of correcting the near-infrared light image before the dark region extraction step, and the direction of the living body region in the near-infrared light image is a predetermined direction. Correction for rotating the near-infrared light image so as to face.

  According to this method, since it can be converted into an image suitable for blood vessel extraction processing for extracting blood vessels, a blood vessel region can be extracted with stable accuracy.

  The image processing apparatus of the present invention irradiates a living body with near-infrared light, picks up the near-infrared wavelength component of reflected light or transmitted light, and digitizes the near-infrared light image relative to peripheral pixels. A dark region extraction unit that extracts a pixel having a small luminance value as a dark pixel and creates dark pixel image data in which a first feature amount is added to each pixel, and a predetermined one or more from the dark pixel image data A continuous dark region extracting unit that extracts the dark pixels continuously distributed so as to match a plurality of mask patterns as a continuous dark region and creates continuous dark region image data in which a second feature amount is written in each pixel; , Extracting a region having the second feature amount larger than a predetermined value from a plurality of continuous dark regions in the continuous dark region image data as a specific region where the blood vessels of the living body exist, and creating specific region image data A specific region extraction unit to perform, Provided.

  According to this configuration, a dark pixel having a luminance value smaller than that of a peripheral pixel in a near-infrared light image is extracted, and further, only a continuous dark pixel is extracted as a continuous dark region from there using a mask pattern, Since the blood vessel region is extracted based on the continuous dark region, the blood vessel region having the intended shape shown in the mask pattern can be accurately extracted with a small amount of calculation. Further, since the calculation amount does not increase or decrease depending on the shape of the blood vessel, stable blood vessel extraction can be performed with a constant processing load and processing time, and the present invention can be applied not only to still images but also to moving images.

  The image processing apparatus of the present invention further includes a calculation region calculation unit that extracts a calculation region image obtained by removing a background region other than the living body from the near-infrared light image, and the dark region extraction unit and the continuous dark region extraction The unit performs processing based on the calculation area image.

  According to this configuration, by setting the calculation region, it is possible to eliminate erroneous extraction of blood vessels in the background region, and to reduce the amount of calculation by narrowing the calculation range.

  Furthermore, in the image processing apparatus of the present invention, an image for correcting the near-infrared light image so that the size of the living body region in the near-infrared light image is a fixed ratio and the direction is in a predetermined direction. A conversion unit is provided.

  According to this configuration, since it can be converted into an image suitable for blood vessel extraction processing for extracting blood vessels, a blood vessel region can be extracted with stable accuracy.

  Furthermore, in the image processing apparatus of the present invention, the dark region extraction unit sequentially extracts a predetermined range of regions while shifting the position by one pixel relative to the near-infrared light image, and extracts at each position. Based on the luminance value of each pixel in the region, a dark pixel evaluation value for evaluating whether the central pixel located at the center of the region is a dark pixel is obtained as the first feature amount, and the obtained dark pixel evaluation The dark pixel image data is created based on the value.

  According to this configuration, it is possible to create dark pixel image data indicating a dark pixel that is a pixel having a luminance value smaller than that of the peripheral region in the near-infrared light image.

  Furthermore, in the image processing apparatus of the present invention, the continuous dark region extracting unit shifts the position of overlapping the mask pattern with respect to the dark pixel image data, while relatively shifting the position of the dark pixel at each position. Continuous evaluation values for evaluating continuity are sequentially obtained, continuous dark regions are extracted based on the obtained continuous evaluation values, and the corresponding dark pixels based on the extracted continuous evaluation values and the position information of the mask pattern The continuous evaluation value is written as the second feature amount to create continuous dark region image data.

  According to this configuration, a continuous dark region that is a dark pixel in which pixels are continuous is extracted from the dark pixel image data using a mask pattern that indicates pixel continuity, and a continuous dark region that indicates the continuous dark region is extracted. Create area image data. Since the dark pixel image data is obtained by extracting pixels that are darker than the surroundings as well as blood vessels, noise other than blood vessels is also included here. Therefore, by extracting dark pixels in which pixels are continuous, that is, dark pixels in a line shape that is a characteristic of blood vessels, it is possible to extract a region having a higher possibility of blood vessels.

  Furthermore, in the image processing device of the present invention, the specific area extraction unit sets, as a specific area, pixels in which the second feature amount is larger than a predetermined threshold at each pixel position of the continuous dark area image data. Extracting and creating the specific area image data based on the second feature amount.

  According to this configuration, a blood vessel region can be extracted by extracting a pixel having a large pixel value from continuous dark region image data. Furthermore, it is possible to know the position where a clearer blood vessel exists based on the size of the pixel value.

  The image processing apparatus according to the present invention further includes an emphasis display unit that uses the specific area image data and emphasizes and displays the specific area in the near-infrared light image according to the second feature amount.

  According to this configuration, since the specific area in the near-infrared light image is displayed with emphasis, it is easy to visually distinguish the specific area.

  According to the present invention, it is possible to accurately extract a blood vessel having an intended shape.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments for carrying out the invention will be described in detail with reference to the drawings.

(Embodiment 1)
As Embodiment 1 of the present invention, a monochrome CCD camera having high near-infrared wavelength sensitivity through an optical filter that irradiates a living body with near-infrared light and transmits reflected light from the living body only through the near-infrared light wavelength. A case where a blood vessel region is extracted from a captured near-infrared light image will be described as an example. In this near-infrared light image, one pixel is represented by 8 bits, and the pixel value is a grayscale image indicating luminance. In the near-infrared light image captured in this way, hemoglobin in the blood has the property of absorbing near-infrared light, so that the reflected light is reduced only in the region where the blood vessel exists, and the pixels of the blood vessel exist. The luminance value is small. Although the reflected light is imaged here, the transmitted light may be imaged, and the luminance value of the pixel in which the blood vessel exists is similarly reduced.

  FIG. 1 is a block diagram showing a schematic configuration of main parts of an image processing apparatus 100 according to Embodiment 1 of the present invention. In the figure, an image processing apparatus 100 according to the first embodiment is formed by a computer, and includes a dark region extraction unit 102, a continuous dark region extraction unit 103, and a specific region extraction unit 104. A near-infrared light image 101 input to the image processing apparatus 100 irradiates a living body with near-infrared light, and captures the reflected light with a monochrome CCD camera through an optical filter that transmits only the near-infrared wavelength. It is a digitized image. A blood vessel region included in the near-infrared light image 101 is extracted as a specific region. The dark region extraction unit 102 extracts dark pixels that are relatively smaller in luminance value than the surrounding pixels in the near-infrared light image 101, creates dark pixel image data indicating the dark pixels, and performs continuous darkness. The data is output to the area extraction unit 103.

  The continuous dark region extraction unit 103 is a dark pixel in which pixels are continuous using a preset mask pattern indicating the pixel continuity with respect to the dark pixel image data output from the dark region extraction unit 102. A continuous dark area is extracted, continuous dark area image data indicating the continuous dark area is created, and output to the specific area extraction unit 104. The specific area extraction unit 104 extracts a specific area including pixels in which blood vessels exist from the continuous dark area image data output from the continuous dark area extraction unit 103, and creates specific area image data indicating the specific area Then, the specific area image data is output as the specific area 105.

Next, the operation of the image processing apparatus 100 configured as described above will be described.
FIG. 2 is a flowchart showing processing performed by the image processing apparatus 100 according to the present embodiment. The flowchart shown in FIG. 2 includes steps S101 to S103, step S101 including steps S1011 to S1015 performs a dark region extraction process, step S102 including steps S1021 to S1027 includes a continuous dark region extraction process, and step S103. Indicates a specific area extraction step.

  That is, the dark region extraction unit 102 performs the process of step S101, the continuous dark region extraction unit 103 performs the process of step S102, and the specific region extraction unit 104 performs the process of step S103, so that the input near-infrared A specific region indicating a blood vessel region in the optical image 101 can be extracted with a small amount of calculation with high accuracy. Furthermore, stable blood vessel extraction can be performed with a constant processing load and processing time, and it can be applied not only to still images but also to moving images.

  Hereinafter, image processing of the image processing apparatus 100 for the near-infrared light image 101 will be described in detail with reference to FIGS. 3 to 7 according to flowcharts.

(Step S101 / Dark region extraction step)
First, in the near-infrared light image 101, a pixel having a luminance value relatively smaller than that of surrounding pixels is extracted as a dark pixel, and multi-value digital image data (darkness) having the same size as the near-infrared light image indicating the dark pixel is extracted. Pixel image data). This is based on the luminance value of each pixel included in the extracted region at each extracted position, by sequentially extracting a certain range of region while relatively shifting the position with respect to the near-infrared light image 101. A dark pixel evaluation value for evaluating whether the central pixel located at the center of the region is a dark pixel is obtained, and dark pixel image data is created based on the obtained dark pixel evaluation value. The dark pixel evaluation value obtained here has such a property that the luminance value of the central pixel is smaller than that of the peripheral pixel, and the larger the luminance value difference, the larger the value. Therefore, it can be said that a pixel having a large dark pixel evaluation value is highly likely to be a blood vessel region that appears dark in the near-infrared light image. In addition, since a local region of a certain range is extracted and the dark pixel evaluation value is obtained in the region, the near infrared light is not uniformly irradiated in the near infrared light image for reasons such as a curved surface of a living body. Even if the brightness of the living body region changes greatly, it is possible to accurately extract a region that is darker than the periphery including the blood vessel region.

  FIG. 3 shows a near-infrared light image obtained by irradiating a near-infrared light source from the upper center of the arm and imaging the arm, where the pixel position crossing the major axis direction of the arm is the horizontal axis, and each pixel position shown on the horizontal axis is It is the figure which represented the luminance value as a vertical axis | shaft. In FIG. 3, the luminance value near the center where the near-infrared light strikes well is high, and the luminance value on the side surface is low. In addition, the luminance value is smaller at the blood vessel position than at the periphery. When the brightness of the living body area changes in this way, a dark area cannot be extracted by binarization processing using a predetermined threshold, but a certain range is cut out and locally darker than surrounding pixels. By examining the pixels, it becomes possible to extract dark pixels with high accuracy.

  Here, the dark pixel extraction processing will be described more specifically. FIG. 4 is a diagram illustrating scanning of dark pixel extraction processing performed on a near-infrared light image. First, the dark pixel image data is initialized with 0 (step S1011), and the first extraction region 401 is set for the near-infrared light image (step S1012). The size of the region to be extracted at this time may be set to a size that always includes the blood vessel region and other living body regions when the region of the blood vessel position is extracted. However, a large size including a plurality of blood vessels is not preferable. From the above, a size about 1.5 times the maximum blood vessel thickness assumed in the near-infrared light image to be captured is set. Here, for explanation, it is assumed that a size of 9 × 9 pixels is set.

  Next, a dark pixel evaluation value at the center pixel 402 located at the center of the extraction region is calculated (step S1013). The dark pixel evaluation value is obtained by obtaining 81 pixels by subtracting the luminance value of the central pixel 402 from the luminance value of each pixel position at each pixel position included in the extraction region 401, and adding all the obtained values. . After the calculation, the extraction area is shifted by one pixel, and the calculation is performed for the next center pixel. The dark pixel evaluation value thus obtained is a positive value when the center pixel is darker than the surrounding pixels, and conversely, a negative value when the center pixel is brighter than the surrounding pixels, and the luminance difference between the center pixel and the surrounding pixels is The larger the value, the larger the absolute value.

  Next, based on the obtained dark pixel evaluation value, a value is written in the pixel position of the dark pixel image data corresponding to the center pixel 402 (step S1014). Note that the dark pixel evaluation value is written when the dark pixel evaluation value is positive, and 0 is written when the dark pixel evaluation value is negative. By doing so, it can be seen that pixels having a pixel value other than 0 in the dark pixel image data are dark pixels. Further, when writing a positive dark pixel evaluation value, the dark pixel evaluation value may be divided by a predetermined value, and the divided value may be written. By performing the division, the value can be reduced while maintaining the characteristics of the dark pixel evaluation value, the image memory size of the dark pixel image data to be secured in advance can be reduced, and the calculation load is reduced in the subsequent processing. Can be reduced. Here, when the dark pixel evaluation value is a positive value, the pixel is a dark pixel. However, a predetermined threshold is set in advance, a pixel having a dark pixel evaluation value larger than the threshold is set as a dark pixel, and the other pixels. May be 0. By setting a value larger than 0 as a threshold value, it is possible to extract only dark pixels that are darker than surrounding pixels.

  Then, the processing from step S1012 to step S1014 is performed on the entire image while shifting the position to be extracted from the near-infrared light image pixel by pixel according to the scanning line 403 shown in FIG. However, the scanning is performed within the scanning range 404 where the extraction region 401 does not protrude from the near-infrared light image (step S1015). By performing the above processing, it is possible to create dark pixel image data indicating a dark pixel that is a pixel having a luminance value smaller than that of the peripheral region in the near-infrared light image. The dark pixel image data indicating the dark pixels created in this way is referred to in the following step S102.

(Step S102 / Continuous Dark Area Extraction Step)
Next, for the dark pixel image data created in step S101, a continuous dark region that is a dark pixel in which pixels are continuous is extracted using a mask pattern that indicates the continuity of pixels set in advance. Then, multi-value digital image data (continuous dark region image data) having the same size as the dark pixel image data indicating the continuous dark region is created. The dark pixel image data created in step S101 is obtained by extracting pixels that are darker than the surroundings as well as blood vessels, and therefore includes noise other than blood vessels. Therefore, by extracting dark pixels in which pixels are continuous, that is, dark pixels in a line shape that is a characteristic of blood vessels, it is possible to extract a region having a higher possibility of blood vessels. This is to sequentially obtain a continuous evaluation value indicating the continuity of dark pixels at each superimposed position using the mask pattern while relatively shifting the position where the mask pattern is superimposed on the dark pixel image data, This is performed by creating continuous dark area image data based on the obtained continuous evaluation value.

  Here, the continuous dark region extraction processing will be described more specifically. FIG. 5 is a diagram showing an example of a mask pattern used in the continuous dark region extraction process. The mask pattern is composed of a first pattern area 501 indicated by diagonal lines and another second pattern area 502. As the image data of 4 × 4 pixels, a pixel indicating the first pattern area is set to 1, 2 The pixel indicating the pattern area is set as 0 in advance. The first pattern area in the mask pattern indicates continuous dark pixels to be extracted, and a set of dark pixels having the shape indicated by the first pattern area is extracted as a continuous dark area.

  First, the continuous dark area image data is initialized with 0 (step S1021). Next, a position for overlaying the mask pattern on the dark pixel image data is set (step S1022). The first set position is the upper left corner of the dark pixel image data as in step S101. Then, at the superimposed position, the mask pattern shown in FIG. 5A is selected as the first mask pattern (step S1023), and the pixel value of the dark pixel image data that overlaps the first pattern area of the selected mask pattern. Based on this, a continuous evaluation value for evaluating the continuity of the pixels having the shape shown in the mask pattern is calculated (step S1024). The continuous evaluation value is obtained by multiplying all the pixel values of the dark pixel image data overlapping the first pattern area. In other words, if the pixel values of the dark pixel image data overlapping the first pattern area are K1, K2, K3, and K4, the continuous evaluation value is K1 × K2 × K3 × K4. By obtaining the continuous evaluation value in this way, the pixel value of the non-dark pixel region in the dark pixel image data is 0, and thus the superimposed dark pixel is indicated in the first pattern region of the mask pattern. If it is not a shape, the continuous evaluation value is zero. Further, in the dark pixel image data, the pixel value is larger as the darker pixel is darker than the surrounding pixels. Therefore, the continuous evaluation value increases as the dark pixels with clearer light and darkness continue. Become.

  Next, based on the continuous evaluation value thus obtained, the continuous evaluation value is written in each pixel position of the continuous dark area image data corresponding to the pixel position indicating the first pattern area of the selected mask pattern (step S1025). Note that the value to be written to the continuous dark area image data may be the obtained continuous evaluation value, but the continuous evaluation value may be divided by a predetermined value and the divided value may be written. By performing division in this way, the value can be reduced while maintaining the characteristics of the continuous evaluation value, the image memory size of the continuous dark area image data secured in advance can be reduced, and computation can be performed in the subsequent processing. The load can be reduced.

  When the writing of the continuous dark area image data in the first mask pattern is finished in this way, the mask pattern shown in FIG. 5B is selected as the second mask pattern, and the first mask pattern is the same as the first mask pattern. In steps S1023 to S1025, the process is repeated (step S1026). This is repeated for all mask patterns shown in FIG. If a value has already been written when writing a value to the continuous dark area image data, the value to be written is compared with the value already written to write a larger value. By doing so, it is possible to prevent overwriting with 0 and to leave a continuous evaluation value indicating the continuity of continuous pixels at the pixel position.

  Also, regarding the process of writing to continuous dark area image data, instead of writing a value for each mask pattern, first, continuous evaluation values for all mask patterns are obtained at the same overlay position, and the maximum continuous The mask pattern having the evaluation value may be examined, and the writing process may be performed only with the mask pattern having the maximum continuous evaluation value. That is, the process of writing the maximum continuous evaluation value only at each pixel position indicated in the first pattern area of the mask pattern having the maximum continuous evaluation value at the overlay position. By this processing, only dark pixels having the largest continuous evaluation value can be extracted in the region where the mask patterns are overlapped, and only a region particularly showing the characteristics of the blood vessel can be extracted.

  When all the mask patterns are completed, the step for superimposing the mask pattern on the dark pixel image data is shifted pixel by pixel according to the scanning line 403 shown in FIG. The processing from S1022 to step S1026 is repeated. The scanning here also prevents the mask pattern from protruding from the dark pixel image data when the mask patterns are overlaid (step S1027). By performing the above processing, continuous dark region image data in which only dark pixels that match the shape shown in the first pattern region of the mask pattern shown in FIG. 5 are extracted from the dark pixels in the dark pixel image data. Can do.

  In this continuous dark region extracting step, a weight value may be given to a preset mask pattern, and the continuous evaluation value obtained for each mask pattern may be weighted. For example, when it is particularly desired to extract dark pixels having the shape shown in FIG. 5B, the mask pattern shown in FIG. 5B is given a weight value having a larger value than other mask patterns, and each mask pattern is set. By multiplying the obtained continuous evaluation value by the weight value and performing weighting, the numerical value of the continuous evaluation value in FIG. 5B can be intentionally increased, and the blood vessels in the long axis direction of the arm are given priority. Contributes to extraction. The weight value used for weighting may be set at a pixel position indicating the first pattern area of the mask pattern. For example, the pixel at the pixel position of the first pattern area of the mask pattern in FIG. When the value is set to 2 and the continuous evaluation value is obtained, the pixel value of the first pattern area is referred to, and the pixel value is used as a weight value to perform weighting, whereby the mask pattern of FIG. The continuous evaluation value can be weighted twice as much as other mask patterns.

  Furthermore, the mask pattern used when obtaining the continuous evaluation value may be changed in accordance with the properties of the near-infrared light image data, particularly the characteristics of the region to be extracted. For example, when it is desired to extract a thick blood vessel, a mask pattern in which the pixel width of the first pattern region is set wide as shown in FIG. 6A is used, and when a long blood vessel is desired to be extracted, as shown in FIG. Such a 5 × 5 pixel image data is prepared, and a mask pattern in which the pixel length of the first pattern region is set long may be used. When there is a traveling direction such as a blood vessel, a mask pattern that matches the traveling direction may be used. For example, when the traveling direction is vertical, only the mask patterns shown in FIGS. 5B, 5C, 5D, 5G, and 5H may be used. By setting the size of the mask pattern and the shape of the first pattern area according to the shape of the area to be extracted in this way, it is possible to easily extract areas of various shapes that are visually easy to understand. . The continuous dark area image data indicating the continuous dark area created in this way is referred to in the following step S103.

(Step S103 / specific area extraction step)
Next, for the continuous dark area image data created in step S102, a specific area composed of pixels in which blood vessels exist is extracted, and the multi-value digital having the same size as the continuous dark area image data is shown. Image data (specific area image data) is created.

  The pixel values of the continuous dark region image data created in step S102 are in the shape of a line that is a characteristic of blood vessels, and the value is larger as the pixel has a smaller luminance value than the surrounding pixels. Therefore, it can be said that the blood vessel region can be extracted by extracting pixels having a large pixel value in the continuous dark region image data. This is done by checking the pixel value of the continuous dark area image data one pixel at a time, comparing the pixel value with a predetermined threshold value at each pixel position, and if it is larger than the threshold value, the pixel of the continuous dark area image data The value is written by writing 0 in other cases, in the corresponding pixel position on the specific area image data. By this processing, it is possible to know the blood vessel region by referring to a pixel whose pixel value is not 0 in the specific region image data. Furthermore, it is possible to know the position where a clearer blood vessel exists based on the size of the pixel value.

  Note that the threshold used here can be set by taking a plurality of near-infrared light images in advance and calculating a value that the observer can best extract using these near-infrared light images. That's fine. For example, when a threshold value with a small value is set, thin blood vessels that are not so dark compared to the surroundings can be extracted, and when a threshold value with a large value is set, only dark blood vessels that are considerably darker than the surroundings can be extracted.

  Further, in setting the threshold value, a threshold value may be calculated and set so that the number of pixels extracted as the specific region is equal to or less than a predetermined number of pixels. For example, the number of pixels in the region of the living body part to be imaged and the ratio of the blood vessel region in the living body part are checked in advance, and the number of pixels is calculated by multiplying the number of pixels of the living body part by the ratio of the blood vessel. A threshold value is set so that the number of pixels in the extracted region is equal to or less than the calculated number of pixels. More specifically, in continuous dark area image data, a histogram is created with the horizontal axis representing the pixel value of the continuous dark area image data and the vertical axis representing the number of pixels having the pixel value. The number of pixels is sequentially accumulated, and a pixel value at a position where the accumulated value exceeds the predetermined number of pixels is set as a threshold value. As a result, even if the brightness of the near-infrared light image picked up for reasons such as how the illumination light strikes, a blood vessel region having a certain number of pixels or more can be extracted constantly.

  Further, in order to increase the accuracy, the following processing may be performed. In the above process, body hair that is close to the characteristics of the blood vessel appears darker than the surroundings in the near-infrared light image, and pixels are continuous, so that it may be erroneously extracted as a specific region indicating the blood vessel. By removing this type of noise, it is possible to further improve the accuracy of blood vessel extraction. This is for the specific area image data generated by the above processing, the feature amount is obtained for each specific area where the pixels are continuous, and the feature amount is included in the specific area for which the feature amount is obtained. After the weighted pixel value is weighted, the hair region is removed by comparing the weighted pixel value with a predetermined threshold again. In this weighting, the pixel value included in the body hair region is intentionally reduced so as to be smaller than the threshold value.

  Specifically, a weighting value represented by a decimal value of 1 or less is calculated according to the feature amount, and weighting processing is performed by multiplying the calculated weighting value by the pixel value of the specific area. Generally, hair is shorter than blood vessels. From this, it can be said that the area value of the blood vessel region is large and the area value of the body hair region is small. Therefore, using the feature amount as the area value of the specific area, in the specific area image data, the area value is obtained for each specific area, and a weight value that decreases as the area value decreases is calculated according to the obtained area value. The pixel values included in the specific area are weighted.

  If there is an area value greater than a preset value, the weight value is set to 1 so that the pixel value is not reduced by weighting. That is, for an area value smaller than a preset value, a weight value represented by a decimal value of 1 or less is calculated so as to be proportional to the area value. The preset value used here may be set by obtaining an area value that can be determined to be a blood vessel from a plurality of near-infrared light images. By this process, the pixel value of the specific area having a small area value can be reduced, and a short area such as body hair can be removed.

  In addition, the area value is calculated by attaching the same label to connected pixels (connected components) and different labels to different connected components, which are generally used when examining the characteristics of a region. This is done by using a labeling process that can separate individual connected components and examine the characteristics of each connected component. First, the specific area image data is labeled and labeled for each specific area. The area value is obtained by counting pixels with the same label in FIG. In addition, among the labeled products, there are cases where a plurality of dark hairs overlap, and in this case, the area value is relatively large compared to simple hairs. In order to remove the hair of the hand, the following weighting process may be further performed after the weighting process.

  FIG. 7 shows an example of a case in which a plurality of blood vessels and dark hair overlap. 7A and 7B are specific regions of blood vessels, and FIG. 7C is a specific region of body hair. A broken line frame in the figure indicates a range surrounded by the vertical width and the horizontal width of the specific area. As shown in FIG. 7A, since the blood vessel has a line-like feature, the difference in the ratio between the vertical width and the horizontal width becomes large. In addition, as shown in FIG. 7B and FIG. 7C, even when the difference in the ratio between the vertical width and the horizontal width is small, when looking at the ratio of the area of the specific region in the range surrounded by the vertical width and the horizontal width, In the case of blood vessels, the ratio is small, and in the case of multiple overlapping hairs, the ratio is large. Therefore, a specific region having a small difference in the ratio between the vertical width and the horizontal width and having a large proportion of the area of the specific region in the range surrounded by the vertical width and the horizontal width is distinguished from body hair, and the pixel value of the distinguished body hair region By intentionally reducing the weight by weighting, the hair region can be removed.

  First, the vertical width and horizontal width of the specific area are obtained. The vertical width and the horizontal width are obtained from the upper, lower, left, and right coordinates located at the end of the pixel having the same label in the labeled specific region. The difference between the ratio of the vertical width and the horizontal width is a division value obtained by dividing the larger value of the vertical width and the horizontal width by the smaller value, and if the obtained division value is equal to or greater than a value that can be determined to be a blood vessel, The value is 1. When the calculated division value is smaller than a value that can be determined to be a blood vessel, the value decreases as the proportion of the area of the specific region in the range surrounded by the vertical and horizontal widths indicated by the broken line in FIG. 7 increases. A weight value is calculated. Specifically, the area value of the specific area is obtained by counting pixels with the same label, the area value of the range surrounded by the vertical width and the horizontal width is obtained by vertical width × horizontal width, and the area value of the specific area Is divided by the area value of the range surrounded by the vertical width and the horizontal width, the weight value is set to 1 when the ratio value is less than or equal to the value that can be determined to be a blood vessel, and it is determined to be a blood vessel For a ratio value larger than the possible value, a weight value represented by a decimal value of 1 or less is calculated so as to be inversely proportional to the ratio value. Note that the values of the division value and the ratio value that can be determined to be necessarily blood vessels may be determined and set in advance by a viewer from a plurality of near-infrared light images.

  As described above, according to the image processing apparatus 100 of the first embodiment, dark pixels having a luminance value smaller than that of the peripheral pixels are extracted from the near-infrared light image 101, and are continuously used from there by using a mask pattern. By extracting only characteristic dark pixels as a continuous dark region and extracting a blood vessel region based on the continuous dark region, it is possible to accurately extract a blood vessel region having the intended shape shown in the mask pattern with a small amount of calculation. . Further, in this processing, the amount of calculation does not increase or decrease depending on the shape of the blood vessel, so that stable blood vessel extraction can be performed with a constant processing load and processing time, and it can be applied not only to still images but also to moving images.

  In the first embodiment, an example in which a blood vessel region is extracted from a near-infrared light image obtained by imaging reflected light obtained by irradiating a living body with near-infrared light has been described. This processing is useful if it is a digital image having a luminance difference between the peripheral region and the peripheral region. For example, the present processing may be performed on an angiographic image obtained by X-ray imaging using an angiographic agent. If the blood vessel region appears brighter than the surrounding region, this processing can be performed on an image obtained by subtracting the pixel value of the captured image from the maximum pixel value that the captured image can take, for example, 255 for an 8-bit digital image. Good. Needless to say, the specific region to be extracted is not limited to the blood vessel region, and is useful even when a specific region having the same characteristics as the blood vessel is extracted.

(Embodiment 2)
Next, an image processing apparatus according to Embodiment 2 of the present invention will be described.
FIG. 8 is a block diagram showing a schematic configuration of a main part of the image processing apparatus according to Embodiment 2 of the present invention. In the figure, an image processing apparatus 800 according to the second embodiment is formed by a computer and includes a calculation area calculation unit 802, a dark area extraction unit 803, a continuous dark area extraction unit 804, and a specific area extraction unit 104. Is done. The same components as those described in the first embodiment are denoted by the same reference numerals, and description thereof is omitted.

  The difference from the configuration of the first embodiment described above is that a calculation area calculation unit 802 is newly added. The calculation area calculation unit 802 calculates a calculation area from the near-infrared light image 101, and calculates the calculated calculation area. The calculation area image data shown is created and output to the dark area extraction unit 803 and the continuous dark area extraction unit 804.

Next, the operation of the image processing apparatus 800 having the above configuration will be described.
FIG. 9 is a flowchart showing processing performed by the image processing apparatus 800 according to the present embodiment. The flowchart of FIG. 9 includes steps S201 to S204, where step S201 is a calculation region calculation step, step S202 consisting of steps S2021 to S2026 is a dark region extraction step, and step S203 consisting of steps S2031 to S2038 is continuous darkness. The area extraction process and step S204 indicate the specific area extraction process. The difference from the processing of the first embodiment is that a calculation area calculation step is newly added, and in the captured digital image, for example, an area excluding a background area other than a living body is calculated as the calculation area, and the calculation indicating the calculated calculation area is performed. Region image data is created, and in the dark region extraction step, the dark region is extracted only in the calculation region. In the continuous dark region extraction step, the continuous dark region is extracted only in the calculation region. The difference is that. By setting the calculation area in this way, it is possible to eliminate erroneous extraction of blood vessels in the background area, and to reduce the calculation amount by narrowing the calculation range.

  That is, the calculation area calculation unit 802 performs the process of step S201, the dark area extraction unit 803 performs the process of step S202, the continuous dark area extraction unit 804 performs the process of step S203, and the specific area extraction unit 104 performs step S204. By performing this process, it is possible to accurately extract a specific region indicating a blood vessel region by eliminating erroneous extraction in the background region, and to reduce the amount of calculation and increase the speed.

  Hereinafter, image processing performed by the image processing apparatus 800 of the present embodiment on a near-infrared light image obtained by irradiating a living body with near-infrared light and capturing reflected light will be described in detail according to a flowchart.

(Step S201 / calculation area calculation step)
First, in the near-infrared light image, a region excluding the background region is calculated as a calculation region, and calculation region image data indicating the calculated calculation region is created. The near-infrared light image captures a living body, but a background region other than the living body is also captured here, and extraction is performed only in the computation region excluding the background region, so that erroneous extraction in the background region is performed. Can be eliminated. In addition, by narrowing down the range to be processed, the number of processes can be reduced and the speed can be increased. The calculation area image data is set by preparing image data of the same size as the near-infrared light image in advance, setting the pixel value of the pixel indicating the calculation area to 1 and setting the others to 0. In the subsequent process, the calculation area can be known by examining the pixel position where the pixel value of the calculation area image data is 1.

  In the calculation area, the luminance value of the near-infrared light image is compared with a predetermined threshold, and a pixel position that satisfies the condition is set. More specifically, when a material that absorbs near-infrared light is used for the background of the living body, the luminance value of the background area in the captured near-infrared light image is smaller than the luminance value of the living body area. FIG. 10 shows a luminance histogram in which a horizontal axis represents a luminance value and a vertical axis represents the number of pixels having the luminance value in a near-infrared light image captured using a material that absorbs near-infrared light on the background of a living body. Is. In FIG. 10, the background region is distributed on the low luminance side, and the living body region is distributed on the high luminance side. Accordingly, a threshold value for separating the distribution of the background region and the living body region is set in advance, and pixels having a luminance value larger than the threshold value are extracted from the near-infrared light image to obtain a calculation region excluding the background region. be able to. Conversely, when a material that reflects near infrared light more strongly than the living body is used for the background of the living body, the calculation is performed by extracting pixels having a luminance value smaller than a predetermined threshold in the near infrared light image. The area can be determined.

  When the brightness changes depending on the amount of irradiated near infrared light, a threshold value to be compared with the brightness value may be calculated based on the brightness value of the near infrared light image and set for each image. This is done by setting the ratio of the background area in the entire image in advance, creating a luminance histogram with the horizontal axis representing the luminance value and the vertical axis representing the number of pixels having that luminance value in the near-infrared light image. The number of pixels is accumulated sequentially from the side, and the luminance value at a position where the accumulated value exceeds the number of pixels in the background area is set as a threshold value. By using the threshold value set in this way, it is possible to obtain a calculation region that constantly has a constant area even when the brightness changes.

  In addition, in the area | region calculated | required in this way, when a some area | region exists, the area | region with the largest area value shall be set as a calculation area | region. FIG. 11 is a diagram in which a region having a luminance value larger than a predetermined threshold is extracted from a near-infrared light image obtained by imaging an arm. In FIG. 11, a region where the luminance value is larger than a predetermined threshold is indicated by hatching. As described above, depending on how the illumination is applied, the background material, and the characteristics of the CCD to be picked up, there may be luminance unevenness and noise in addition to the biological region originally set as the calculation region. However, the area other than the living body area is generally smaller than the living body area. Thus, by setting only a region having the maximum area value as a calculation region, a calculation region including only a living body region can be obtained. As in the first embodiment, the area value is obtained by performing a labeling process on the calculation area, and the pixel value of the area having a label other than the label attached to the area having the maximum area value. Is set to 0.

  In addition, although it is possible to set the calculation area with high accuracy by performing the above processing, it is also possible to perform a modification that contracts the size of the calculation area with respect to the set calculation area. When imaging is performed by irradiating the arm with near-infrared light, the side surface of the arm is not sufficiently irradiated with near-infrared light, and blood vessel characteristics do not appear. Further, when the inside of the arm is imaged, there is body hair near the side surface, so it is considered that the vicinity of the side surface should be removed from the calculation area. Accordingly, the side pixels are excluded from the calculation area by shrinking the calculation area inward by several pixels. Shrinkage processing is performed by setting a pixel that has a pixel value of 1 in the calculation area to 0 if there is even one in the vicinity, and the size of the original calculation area is 10% or less smaller. Repeat the contraction until The above processing is performed to obtain a calculation area, and the calculated calculation area image data is referred to in steps S202 and S203 below.

(Step S202 / Dark Region Extraction Step)
Next, as in step S101 in FIG. 2, in the near-infrared light image, a dark pixel that is a pixel having a relatively smaller luminance value than the surrounding pixels is extracted, and the near-infrared light image indicating the dark pixel is extracted. Multi-valued digital image data (dark pixel image data) of the same size as is created. However, the process of extracting dark pixels is performed only in the calculation area obtained in step S201.

  First, the dark pixel image data is initialized with 0 (step S2021), and the first extraction region 401 is set for the near-infrared light image as shown in FIG. 4 (step S2022). Next, it is determined whether or not the central pixel 402 positioned at the center of the extraction area is the calculation area (step S2023). If it is the calculation region, the dark pixel evaluation value is calculated (step S2024) and dark pixel image data is created (step S2025), as in step S101 of FIG. If it is not the calculation area, step S2024 and step S2025 are skipped. Then, the processing from step S2022 to step S2025 is performed on the entire image while shifting the position to be extracted from the near-infrared light image by one pixel according to the scanning line 403 shown in FIG. 4 (step S2026). The dark pixel image data indicating the dark pixels created in this way is referred to in the following step S203.

(Step S203 / Continuous Dark Area Extraction Step)
Next, as in step S102 of FIG. 2, the dark pixels in which the pixels are continuous using the preset mask pattern indicating the continuity of the pixels for the dark pixel image data created in step S202. Are extracted, and multi-valued digital image data (continuous dark region image data) having the same size as the dark pixel image data indicating the continuous dark region is created. However, the process of extracting the continuous dark area is performed only in the calculation area obtained in step S201.

  First, as in step S102, the continuous dark area image data is initialized with 0 (step S2031). Next, a position for superimposing the mask pattern on the upper left corner of the dark pixel image data is set (step S2032). Then, it is determined whether or not it is a calculation area at the overlapping position (step S2033). When the pixels shown in the first pattern area of all the mask patterns to be superimposed are included in the calculation area, it is determined that the calculation area. If it is the calculation area, the mask pattern is selected (step S2034), the continuous evaluation value is calculated (step S2035), and the continuous dark area image data is created (step S2036), as in step S102 of FIG. ) Is repeated for all mask patterns (step S2037). If it is not the calculation area, steps S2034 to S2037 are skipped. Then, the processing from step S2032 to step S2037 is performed on the entire image while shifting the position of overlapping the dark pixel image data one pixel at a time according to the scanning line 403 shown in FIG. 4 (step S2038). The continuous dark area image data indicating the continuous dark area thus created is referred to in the following step S204.

(Step S204 / specific area extraction step)
Next, for the continuous dark area image data created in step S203, a specific area composed of pixels in which blood vessels exist is extracted, and the multi-value digital having the same size as the continuous dark area image data indicating the specific area is extracted. Image data (specific area image data) is created. The method for extracting the specific area and creating the specific area image data is the same as that in step S103 in FIG. The specific area image data thus created is output as a specific area where a blood vessel exists.

  As described above, according to the image processing apparatus 800 of the second embodiment, the calculation region is set, and only in the calculation region, the dark region extraction process or the continuous dark region extraction process is performed, thereby the background region. It is possible to accurately extract a specific region indicating a blood vessel region by eliminating erroneous extraction, and to reduce the amount of calculation and increase the speed.

  In the second embodiment, an example in which a blood vessel region is extracted from a near-infrared light image obtained by imaging reflected light obtained by irradiating a living body with near-infrared light has been described. It is useful if it is a digital image that has a luminance difference between the peripheral region and the peripheral region. Needless to say, the specific region to be extracted is not limited to the blood vessel region, and is useful even when a specific region having the same characteristics as the blood vessel is extracted.

  (Embodiment 3)

Next, an image processing apparatus according to Embodiment 3 of the present invention will be described.
FIG. 12 is a block diagram showing a schematic configuration of a main part of an image processing apparatus according to Embodiment 3 of the present invention. In the figure, an image processing apparatus 1200 according to the third embodiment is formed by a computer and includes an image conversion unit 1201, a dark region extraction unit 102, a continuous dark region extraction unit 103, and a specific region extraction unit 104. Is done. The same components as those described in the first embodiment are denoted by the same reference numerals and description thereof is omitted.

  The difference from the configuration of the first embodiment is that an image conversion unit 1201 is newly added. The image conversion unit 1201 converts the near-infrared light image 101, and the converted near-infrared light image is sent to the dark region extraction unit 102. It is a point to output.

  Using the image processing apparatus 1200 configured as described above, the image conversion unit 1201 performs the processing performed in step S301, the dark region extraction unit 102 performs the processing in step S302, and the continuous dark region extraction unit 103 performs the processing in step S303. By performing the processing performed in step S304 in the specific region extraction unit 104, it is not necessary to change parameters in each step due to a change in the captured near-infrared light image, and the specific region indicating the blood vessel region with stable accuracy Can be extracted.

  Hereinafter, image processing performed by the image processing apparatus 1200 according to the present embodiment on a near-infrared light image obtained by irradiating a living body with near-infrared light and capturing reflected light in the same manner as in the first embodiment will be described in detail with reference to the flowcharts. To do.

  FIG. 13 is a flowchart of image processing performed by the image processing apparatus 1200 according to the third embodiment. The flowchart shown in the figure includes steps S301 to S304, where step S301 is an image conversion step, step S302 consisting of steps S3021 to S3025 is a dark region extraction step, and step S303 consisting of steps S3031 to S3037 is continuous. The dark region extraction step and step S304 indicate the specific region extraction step, respectively. The difference from the processing of the first embodiment described above is that a specific region such as the size and direction of a blood vessel in the digital image is added so that a new image conversion process is added and it is not necessary to set parameters for each captured digital image. The difference is that the digital image is converted so that the information is constant, and in the dark region extraction step, the dark region is extracted from the converted digital image.

(Step S301 / Image Conversion Step)
First, in the captured near-infrared light image, the near-infrared light image is converted so that information on a specific region such as the size and direction of a blood vessel in the image is constant. The size of the living body displayed in the near-infrared light image changes sequentially due to the distance variation between the living body and the CCD camera. In addition, the direction of the living body displayed in the near-infrared light image changes sequentially as the living body moves. Accordingly, in accordance with the captured near-infrared light image, the size and direction of the mask pattern to be overlaid in the dark region extraction process described in the first and second embodiments, the extracted region range, and the continuous dark region process, It is also necessary to change various parameters such as threshold values for each process. In this case, there are many parameters to be changed, which is very time-consuming, and the accuracy of blood vessel extraction may change for each image. Therefore, the captured near-infrared light image is converted into an image suitable for blood vessel extraction processing for extracting blood vessels, and blood vessel extraction processing is performed on the converted near-infrared light image. This eliminates the need to change the parameters used in each process.

  Hereinafter, an example of image conversion in a near-infrared light image obtained by imaging an arm will be shown. First, the size of the near-infrared light image is changed so that the size of the blood vessel is always constant. This is done by using the area value of the arm region shown in the near-infrared light image, and the area value of the reference arm region is set first, and this reference area value and the newly captured near area are set. Conversion is performed to enlarge or reduce the image size at the same magnification in the horizontal and vertical directions so that the area values of the arm regions of the infrared light image are equal. The area value of the arm region is obtained by extracting the arm region and counting the number of pixels in the extracted region by the same process as the process of obtaining the arm area excluding the background as the calculation area in step S201 of the second embodiment. Can be sought.

The near-infrared light image is also rotated so that the direction of the blood vessel is always constant. The blood vessels of the arm run in the longitudinal direction of the arm. Therefore, similarly, the arm region is extracted, a straight line indicating the longitudinal direction of the arm is obtained, and the near-infrared light image is rotated so that the obtained straight line matches a preset reference straight line. As shown in FIG. 14, when the horizontal direction of the near-infrared light image is set as the X axis and the vertical direction is set as the Y axis, the straight line indicating the longitudinal direction of the arm is shifted by one pixel from the upper part of the arm region in the Y axis direction. Then, the midpoint of the arm region in the X-axis direction is obtained sequentially, and the obtained midpoint of the arm region in each Y-axis is obtained as a straight line connecting by least square approximation. Note that image enlargement, reduction, and rotation may be performed using two-dimensional affine transformation, which is generally used. In the affine transformation, the coordinates before transformation are (x, y), the coordinates after transformation are (x ′, y ′), the counterclockwise rotation angle is θ, and the scaling factor is s as follows: expressed.
x ′ = (x × cos θ + y × sin θ) × s
y ′ = (− x × sin θ + y × cos θ) × s

The near-infrared light image thus converted is referred to in the following step S302.
(Step S302 / Dark Region Extraction Step)

  Next, in the same manner as in step S101 in FIG. 2, dark pixels that are relatively smaller in luminance value than peripheral pixels are extracted from the near-infrared light image converted in step S301, and the dark pixels are extracted. Multi-value digital image data (dark pixel image data) having the same size as that of the near-infrared light image indicating the pixel is created. The method for extracting dark pixels and creating dark pixel image data is the same as that in step S101 in FIG. The dark pixel image data indicating the dark pixels created in this way is referred to in the following step S303.

(Step S303 / Continuous Dark Area Extraction Step)
Next, in the same manner as in step S102 of FIG. 2, the dark pixels in which the pixels are continuous using the preset mask pattern indicating the continuity of the pixels for the dark pixel image data generated in step S302. Are extracted, and multi-valued digital image data (continuous dark region image data) having the same size as the dark pixel image data indicating the continuous dark region is created. Note that the method of extracting continuous dark regions and creating continuous dark region image data is the same as step S102 in FIG. The continuous dark area image data indicating the continuous dark area created in this way is referred to in the following step S304. (Step S304 / specific dark region extraction step)

  Next, in the same manner as in step S103 of FIG. 2, the continuous dark region image data indicating the specific region is extracted from the continuous dark region image data created in step S303 by extracting a specific region including pixels in which blood vessels exist. Multi-valued digital image data (specific area image data) of the same size as is created. The method for extracting the specific area and creating the specific area image data is the same as that in step S103 in FIG. The specific area image data thus created is output as a specific area where a blood vessel exists.

  As described above, according to the image processing apparatus 1200 of the third embodiment, the near-infrared light image is converted so that the information on the specific region in the captured near-infrared light image is constant, and the dark region extraction step is performed. Then, by performing the process of extracting dark pixels from the converted near-infrared light image, there is no need to change parameters in each step due to changes in the captured near-infrared light image, and blood vessel regions with stable accuracy Can be extracted.

  In Embodiment 3, an example in which a blood vessel region is extracted from a near-infrared light image obtained by imaging reflected light obtained by irradiating a living body with near-infrared light has been described. It is useful if it is a digital image that has a luminance difference between the peripheral region and the peripheral region. Needless to say, the specific region to be extracted is not limited to the blood vessel region, and is useful even when a specific region having the same characteristics as the blood vessel is extracted.

  Further, the image conversion process described in the third embodiment may be added to the image processing method according to the second embodiment. Specifically, first, image conversion shown in step S301 is performed on the captured near-infrared light image, and processing in steps S201 to S204 is performed on the converted near-infrared light image. Thereby, both effects shown in the second embodiment and the third embodiment can be obtained.

(Embodiment 4)
Next, an image processing apparatus according to the fourth embodiment in which the image conversion process in the image processing apparatus 1200 according to the third embodiment is added to the image processing in the image processing apparatus 1100 according to the second embodiment will be described.

  FIG. 15 is a block diagram showing a schematic configuration of main parts of an image processing apparatus 1500 according to Embodiment 4 of the present invention. In the figure, an image processing apparatus 1500 according to the fourth embodiment is formed by a computer, and includes an image conversion unit 1201, a calculation region calculation unit 802, a dark region extraction unit 803, a continuous dark region extraction unit 804, and a specific region extraction unit. 104 is comprised. Components having the same configurations as those described in Embodiment 2 and FIG. 12 are denoted by the same reference numerals, and description thereof is omitted. The difference from the configuration of the second embodiment and FIG. 12 is that the image conversion unit 1201 converts the near-infrared light image 101 and outputs the converted near-infrared light image to the calculation region calculation unit 802 and the dark region setting unit 803. It is a point to do.

  Using the image processing apparatus 1500 configured as described above, the image conversion unit 1201 performs the processing performed in step S301, the calculation region calculation unit 802 performs the processing performed in step S201, and the dark region extraction unit 803 performs the processing performed in step S202. The continuous dark region extraction unit 804 performs the processing performed in step S203, and the specific region extraction unit 104 performs the processing performed in step S204, thereby obtaining both effects of the second embodiment and the third embodiment. Can do.

(Embodiment 5)
Next, an image processing apparatus according to Embodiment 5 of the present invention will be described with reference to the drawings. The image processing apparatus according to the fifth embodiment uses the specific area image data output from the image processing apparatus 100 according to the first embodiment to highlight and display the specific area indicating the blood vessel area in the near-infrared light image. An image processing apparatus.

  FIG. 16 is a block diagram illustrating a schematic configuration of main parts of an image processing apparatus 1600 according to the present embodiment. In the figure, an image processing apparatus 1600 according to the present embodiment uses the image processing apparatus 100 that creates specific area image data indicating blood vessels extracted from the near-infrared light image according to the first embodiment. The image processing apparatus 100 and a highlight display unit 1601 that highlights and displays a specific area indicating a blood vessel area in the near-infrared light image 101 according to the size of the pixel value of the specific area image data output from the image processing apparatus 100. It is comprised including. In addition, the same code | symbol is attached | subjected about the thing of the same structure demonstrated in Embodiment 1, and description is abbreviate | omitted. The image processing device 1600 creates specific region image data indicating a blood vessel region in the image processing device 100 with respect to the captured near-infrared light image 101, and the highlighting display unit 1601 displays the specific region image in the near-infrared light image 101. An image in which the specific area indicated in the data is emphasized is created, and the created image is displayed as the emphasized image 1602.

  According to the image processing apparatus 1600 configured as described above, it is possible to emphasize and display a specific area in a near-infrared light image, and to easily distinguish the specific area visually.

  Hereinafter, the process performed in the highlight display unit 1601 will be described in detail. The specific area image data output from the image processing apparatus 100 has a larger value as the pixel more strongly represents the characteristics of the blood vessel. Therefore, the larger the pixel value indicated in the specific area image data is, the stronger the process of enhancing the corresponding pixels of the near-infrared light image 101 is. Further, as a method for enhancing an image, a process of correcting color information in a specific area is performed. This is a gray-scale image in which the pixel value of the near-infrared light image indicates luminance, but the emphasized image 1602 to be emphasized and displayed is a color image in which one pixel has a luminance value of 8 bits for each RGB, and the near-infrared light image This is done by coloring the pixels in the specific area. The correction of the color information is a correction for emphasizing the red component. For pixels that are not the specific area, the luminance value of the near-infrared light image is substituted for each of the RGB values, and for the pixels that are the specific area, the R value is set. Is set to a relatively large value relative to the G value and the B value, the R value represents the brightness value of the near-infrared light image, and the G value and B value represent the brightness value of the near-infrared light image. This is done by substituting a value obtained by subtracting the pixel value. If the subtraction result is negative, 0 is substituted.

  By performing the above processing, as the pixel value of the specific area image data increases, the R value becomes relatively larger than the G value and the B value, and the red component can be strongly emphasized. Since the enhancement process using the luminance value of the image is performed, the red component of the specific region indicating the blood vessel region can be emphasized so that the boundary is not unnatural by using the luminance value of the specific region. In addition, the intensity of image enhancement can be adjusted by increasing or decreasing the pixel value of the specific area image data to be subtracted from the luminance value of the near-infrared light image at a certain ratio in the G value and the B value. . When the pixel value of the specific area image data is increased, the red component is corrected more strongly, and when the pixel value of the specific area image data is decreased, the red component is corrected weakly.

  Note that the method shown here is merely an example, and it goes without saying that there is no problem even if the specific region is emphasized using the specific region image data by another method. For example, as another method for emphasizing an image, in the case of a near-infrared light image obtained by imaging a living body, the living body region may be displayed in skin color, and correction may be performed to emphasize the green component in a specific region indicating a blood vessel. . Thereby, the near-infrared light image which is a grayscale image can be displayed as a color image. Specifically, the RGB value of each pixel in the enhanced image 1602 is obtained by an expression represented by R = Y + Cr, G = Y−0.51 × Cr−0.186 × Cb, and B = Y + Cb, where Y is The brightness value of the near-infrared light image, Cr is 40, Cb is -64, and the biological region is colored with skin color. In the specific region indicating the blood vessel, the specific region image is obtained from the R value of the obtained RGB value. By subtracting the pixel value of the data from the G value by dividing the pixel value of the specific area image data by 16 and subtracting the value of the pixel value of the specific area image data by 4 from the B value, the skin color The green component can be emphasized in a specific region indicating a blood vessel on the living body region colored in (1). Note that the formulas and numerical values shown here are examples, and the color may be adjusted by changing the values.

  As described above, according to the image processing apparatus 1600 according to Embodiment 5, a blood vessel region can be extracted from a near-infrared light image, the extracted blood vessel region can be emphasized without a sense of incongruity, and can be easily visually distinguished. can do.

  In the fifth embodiment, the apparatus for extracting the blood vessel region from the near-infrared light image and creating the specific region image data indicating the blood vessel region has been described as the image processing device 100 shown in FIG. The image processing apparatus 800 shown in FIG. 12, the image processing apparatus 1200 shown in FIG. 12, or the image processing apparatus 1500 shown in FIG. Thereby, it is possible to perform the enhancement process using the specific area image data shown in the second or third embodiment. In Embodiments 1 to 4, reflected light is imaged. However, transmitted light may be imaged, and similarly, the luminance value of a pixel in which a blood vessel exists is reduced. By doing so, blood vessel extraction can be performed.

  Although various embodiments of the present invention have been described above, the present invention is not limited to the matters shown in the above-described embodiments, and those skilled in the art can make modifications and applications based on the description and well-known techniques. This is also the scope of the present invention, and is included in the scope for which protection is sought.

  INDUSTRIAL APPLICABILITY The present invention has an effect that blood vessels having intended shape characteristics can be accurately extracted, and various processing devices for extracting a specific region in a digital image, particularly shape characteristics having pixel continuity. The present invention can be applied to an image processing apparatus that extracts a target object.

1 is a block diagram showing a schematic configuration of a main part of an image processing apparatus according to Embodiment 1 of the present invention. The flowchart explaining operation | movement of the image processing apparatus which concerns on Embodiment 1 of this invention. The figure which shows the luminance value of the arm pixel in the near-infrared light image at the time of imaging an arm in the image processing apparatus which concerns on Embodiment 1 of this invention. The figure which shows the scanning of the dark area extraction process which the image processing apparatus which concerns on Embodiment 1 of this invention performs with respect to a near-infrared light image The figure which shows an example of the mask pattern used at the continuous dark area | region extraction process of the image processing apparatus which concerns on Embodiment 1 of this invention. The figure which shows an example of a deformation | transformation of the mask pattern used at the continuous dark area | region extraction process of the image processing apparatus which concerns on Embodiment 1 of this invention. The figure which shows an example of the specific area | region of the case where the blood vessel and dark body hair have overlapped in the image processing apparatus which concerns on Embodiment 1 of this invention. FIG. 3 is a block diagram showing a schematic configuration of a main part of an image processing apparatus according to Embodiment 2 of the present invention. The flowchart explaining operation | movement of the image processing apparatus which concerns on Embodiment 2 of this invention. The figure which shows the luminance histogram of the near-infrared-light image imaged using the material which absorbs near-infrared light in the background in the image processing apparatus which concerns on Embodiment 2 of this invention. The figure which shows the area | region which has a luminance value larger than a predetermined threshold in the image processing apparatus which concerns on Embodiment 2 of this invention. FIG. 3 is a block diagram showing a schematic configuration of a main part of an image processing apparatus according to Embodiment 3 of the present invention. The flowchart explaining operation | movement of the image processing apparatus which concerns on Embodiment 3 of this invention. The figure explaining the method of calculating | requiring the straight line which shows the longitudinal direction of an arm in the image processing apparatus which concerns on Embodiment 3 of this invention. FIG. 6 is a block diagram showing a schematic configuration of a main part of an image processing apparatus according to Embodiment 4 of the present invention. FIG. 6 is a block diagram showing a schematic configuration of a main part of an image processing apparatus according to Embodiment 5 of the present invention. Flowchart explaining conventional blood vessel extraction processing

Explanation of symbols

100, 800, 1200, 1500, 1600 Image processing device 101 Near-infrared light image 102 Dark region extraction unit 103 Continuous dark region extraction unit 104 Specific region extraction unit 105 Specific region 401 First extraction region 402 Located at the center of the extraction region Center pixel 403 Extraction processing scanning line 404 Extraction processing scanning range 501 First pattern region 502 Second pattern region 802 Calculation region calculation unit 803 Dark region extraction unit 804 Continuous dark region extraction unit 1201 Image conversion unit 1601 Highlight display unit 1602 Enhanced image

Claims (25)

A near-infrared light image that is irradiated with near-infrared light and picks up the near-infrared wavelength component of reflected or transmitted light and digitizes it, darkens pixels that have relatively lower luminance values than the surrounding pixels. A dark region extracting step of creating dark pixel image data that is extracted as pixels and added with a first feature amount to each pixel;
From the dark pixel image data created in the dark region extraction step, the dark pixels that are continuously distributed so as to match one or a plurality of mask patterns set in advance are extracted as continuous dark regions. A continuous dark area extracting step of creating continuous dark area image data in which the feature amount of 2 is written;
From the plurality of continuous dark areas in the continuous dark area image data created in the continuous dark area extraction step, an area where the second feature amount is larger than a predetermined value is extracted as a specific area where a blood vessel of the living body exists. And a specific area extraction step of creating specific area image data,
An image processing method comprising: The image processing method according to claim 1,
The dark region extracting step sequentially extracts a predetermined range of regions while shifting the position of each pixel relative to the near-infrared light image, and based on the luminance value of each pixel in the region extracted at each position. In addition, a dark pixel evaluation value for evaluating whether or not a central pixel located at the center of the region is a dark pixel is obtained as the first feature amount, and the dark pixel image data is obtained based on the obtained dark pixel evaluation value. Image processing method to create. The image processing method according to claim 2,
The dark region extraction step obtains, for each pixel, a value obtained by subtracting the luminance value of the central pixel from the luminance value of all pixels in the extracted region, and sums the values of all pixels after the subtraction An image processing method. The image processing method according to claim 2,
In the dark region extraction step, a pixel in which the dark pixel evaluation value of the central pixel is larger than a predetermined threshold is set as the dark pixel, and in the extracted predetermined region, the central pixel is a dark pixel. The result obtained by dividing the dark pixel evaluation value of the dark pixel by a predetermined value is added as a feature amount of the dark pixel, all pixels other than the dark pixel are added as a feature amount, and the center pixel is a dark pixel. 3. The image processing method according to claim 2, wherein zero is added as a feature amount to all pixels when there is no image. The image processing method according to claim 1,
In the continuous dark region extraction step, successive evaluation values for evaluating the continuity of dark pixels at each position are sequentially shifted while the position where the mask pattern is superimposed is relatively shifted by one pixel with respect to the dark pixel image data. Obtaining and extracting a continuous dark region based on the continuous evaluation value, and writing the continuous evaluation value as the second feature amount in a corresponding dark pixel based on the continuous evaluation value and the position information of the mask pattern An image processing method for creating continuous dark area image data by using the above method. The image processing method according to claim 5,
The continuous dark region extracting step is an image processing apparatus that assigns a different weight value to each mask pattern and weights a continuous evaluation value obtained for each mask pattern. The image processing method according to claim 5,
In the continuous dark region extraction step, the continuous evaluation value is divided by a predetermined value at a pixel position on the continuous dark region image data corresponding to a position matching the mask pattern in the generation of the continuous dark region image data. An image processing method in which a result is written as the second feature amount, and if a value has already been written, writing is performed as the second feature amount only when the value to be written is a large value. The image processing method according to claim 5,
In the continuous dark region extraction step, in the creation of the continuous dark region image data, the combination of the continuous dark region having the maximum continuous evaluation value among the continuous evaluation values and the mask pattern is checked, and only the corresponding dark pixel is checked. An image processing method in which the continuous evaluation value is written as the second feature amount. The image processing method according to claim 1,
The specific region extracting step extracts, as a specific region, a pixel having the second feature amount larger than a predetermined threshold at each pixel position of the continuous dark region image data, and the second feature amount is extracted. An image processing method for creating specific area image data based on the image data. The image processing method according to claim 9, comprising:
The specific area extracting step calculates, as the predetermined threshold, a value such that the number of pixels extracted as the specific area from the continuous dark area image data is equal to or less than a predetermined number of pixels, and sets the value Image processing method. The image processing method according to claim 9, comprising:
In the specific area extraction step, after the generation of the specific area image data, weighting is performed on each pixel for which a weight value is calculated according to an area value of a continuous dark area or a vertical / horizontal width ratio. An image processing method in which only a pixel area whose value is larger than a predetermined threshold is extracted again as a specific area. The image processing method according to claim 1,
Before the dark region extraction step, it comprises a calculation region calculation step of extracting a calculation region image excluding the background region other than the living body from the near-infrared light image, and in the dark region extraction step and the continuous dark region extraction step, An image processing method for performing processing based on the calculation area image. The image processing method according to claim 12,
The calculation region calculation step is an image processing method for extracting a calculation region by comparing a luminance value of each pixel of the near-infrared light image with a predetermined threshold value and generating the calculation region image. The image processing method according to claim 13,
The calculation region calculation step is an image processing method in which the predetermined threshold is set to a value such that the number of pixels extracted from the near-infrared light image is equal to or greater than a predetermined number of pixels. The image processing method according to claim 13,
In the calculation area calculation step, when there are a plurality of areas in which pixels that are candidates for the calculation area are collected, the image processing method uses the area having the maximum number of pixels as the calculation area image. . The image processing method according to claim 13,
The calculation area calculation step is an image processing method for performing a correction to reduce the size of the calculation area with respect to the extracted calculation area. The image processing method according to claim 1,
An image conversion step for correcting the near-infrared light image before the dark region extraction step, and the near-infrared light so that the size of the living body region in the near-infrared light image is a constant ratio An image processing method for performing correction for enlarging or reducing the size of an image. The image processing method according to claim 1,
An image conversion step for correcting the near-infrared light image before the dark region extraction step, and the near-infrared light image so that a direction of a living body region in the near-infrared light image is oriented in a predetermined direction. Image processing method for performing correction to rotate the image. A near-infrared light image that is irradiated with near-infrared light and picks up the near-infrared wavelength component of reflected or transmitted light and digitizes it, darkens pixels that have relatively lower luminance values than the surrounding pixels. A dark area extraction unit that extracts dark pixel image data that is extracted as pixels and adds a first feature amount to each pixel;
From the dark pixel image data, the dark pixels continuously distributed so as to match one or a plurality of mask patterns set in advance are extracted as continuous dark regions, and a second feature amount is written to each pixel. A continuous dark area extraction unit for creating dark area image data;
An area having the second feature amount larger than a predetermined value is extracted as a specific area where the blood vessels of the living body are present from a plurality of continuous dark areas in the continuous dark area image data, and specific area image data is generated. A specific area extraction unit;
An image processing apparatus comprising: The image processing apparatus according to claim 19,
A calculation area calculation unit that extracts a calculation area image excluding a background area other than the living body from the near-infrared light image,
The dark region extraction unit and the continuous dark region extraction unit are image processing devices that perform processing based on the calculation region image. The image processing apparatus according to claim 19,
An image processing apparatus comprising: an image conversion unit that corrects the near-infrared light image so that a size of a living body region in the near-infrared light image is a constant ratio and a direction is in a predetermined direction. The image processing apparatus according to claim 19,
The dark region extraction unit sequentially extracts a predetermined range of regions while shifting the position by one pixel relative to the near-infrared light image, and based on the luminance value of each pixel in the region extracted at each position. In addition, a dark pixel evaluation value for evaluating whether or not a central pixel located at the center of the region is a dark pixel is obtained as the first feature amount, and the dark pixel image data is obtained based on the obtained dark pixel evaluation value. An image processing device to be created. The image processing apparatus according to claim 19,
The continuous dark region extraction unit sequentially shifts the evaluation value for evaluating the continuity of dark pixels at each position while relatively shifting the mask pattern overlapping position with respect to the dark pixel image data by one pixel at a time. A continuous dark region is extracted based on the continuous evaluation value, and the continuous feature value is extracted from the extracted continuous evaluation value and the position information of the mask pattern. An image processing apparatus for creating continuous dark area image data by writing as The image processing apparatus according to claim 19,
The specific area extraction unit extracts, as a specific area, a pixel in which the second feature amount is larger than a predetermined threshold at each pixel position of the continuous dark region image data, and extracts the second feature amount. An image processing apparatus for creating the specific area image data based on the image data. The image processing apparatus according to claim 19,
An image processing apparatus comprising: an emphasis display unit that uses the specific area image data and emphasizes and displays the specific area in the near-infrared light image according to the second feature amount.

Images