JP2006293522A - Straight line detection device, straight line detection method, and program for the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To detect straight lines even if points on the straight lines deviate largely. <P>SOLUTION: Candidate points on straight lines in a real space are detected from image data, and coordinate values of the candidate points are projected on a Hough space to produce curves LS on the Hough space. A plurality of detection intervals M each grouping a plurality of subintervals ΔD obtained by divinding the Hough space in a predetermined size are set, and curves LS passing each detection interval M are counted. The straight lines on the real space are computed from the coordinate values of any subintervals ΔD in detection intervals M in which the number of curves LS passing the detection interval M exceeds a first threshold. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an irradiation field determination device that recognizes an irradiation field of a radiographic image, an irradiation field determination method, and a program thereof.

  Conventionally, there are various methods of reading a recorded radiographic image to obtain image data, performing appropriate image processing on the image data, and then reproducing a visible image suitable for interpretation based on the processed image data. In the field.

  By the way, when radiographic images are captured and recorded in X-ray imaging, in order to minimize the influence on the living body due to radiation irradiation, and to prevent deterioration in image quality performance due to scattered light from parts unnecessary for observation, etc. In many cases, an irradiation field stop made of lead or the like is used to limit the irradiation area so that the radiation is irradiated only to a necessary part of the subject.

  When photographing is performed using an irradiation field stop, an image of the subject or the like is recorded in an internal region (irradiation field region) of the opening contour of the irradiation field stop on a recording medium such as a stimulable phosphor sheet. Radiation does not reach the outer region (irradiation field region) and is in an unexposed state. That is, the irradiation field contour of the image corresponding to the opening contour is an edge line.

  Then, when image processing is performed by reading image data from a recording medium in which an image is recorded only in the irradiation field region in this way, if gradation processing or the like is performed only on the image data in the irradiation field region, The number of processes can be greatly reduced, and the processing load can be reduced and the processing speed can be improved.

  On the other hand, since the irradiation field area is in an unexposed state, in a negative image such as a medical X-ray film, it becomes the lowest density area (area having a low pixel value). When observing a transmitted image with light, the minimum density area is very bright, so that the brightness of the irradiation field area, especially the area close to the irradiation field area, is affected. Interpretation performance is reduced. Similarly, when the image is displayed on an image display device such as a CRT, the irradiation field region has a high luminance, which causes a problem in interpretation of the image in the irradiation field.

  Therefore, in the radiographic image recording / reproducing system, processing for forcibly replacing each image data for such an irradiation field region uniformly with a value corresponding to the highest density (or lowest luminance) is performed. This process is generally called a blackening process. To perform this blackening process, it is very important to accurately recognize the irradiation field contour.

  For example, it is well known to obtain an irradiation field contour by searching for a portion where the change in image data is steep, utilizing the fact that the irradiation field contour becomes an edge line in which the density change of the image changes sharply. However, as a specific method for obtaining the edge line, a plurality of radial straight lines from a predetermined point of the image (for example, the center point of the image) to the edge of the image are set, and the direction of each of these straight lines is set. A method has been proposed in which candidate points on an edge having a large difference in data for each direction are detected based on the image data along the line, and a line serving as an edge is detected based on these candidate points (for example, Patent Documents). 1).

Alternatively, a method has been proposed in which a straight line is obtained by applying a Hough transform to these edge candidate points, and a region surrounded by the straight line is set as an irradiation field region (for example, Patent Document 2).
Japanese Unexamined Patent Publication No. 63-259538 JP-A-10-275213

  However, when a straight line is detected by using the Hough transform based on the candidate points on the edge obtained along the radial direction described above, the candidate points on the edge are not necessarily in a straight line. It is not necessarily lined up, and there is some variation in width. For this reason, even if many points serving as edges are detected, a correct straight line may not be detected if the variation is large.

  The present invention has been made in view of the above circumstances, and it is an object of the present invention to provide a straight line detection device, a straight line detection method, and a program therefor, in which the straight line detection probability is improved over conventional straight line detection processing. .

The straight line detection device of the present invention includes an image data input means for inputting image data,
Candidate point detecting means for detecting a plurality of candidate points on a straight line in real space from the input image data;
Projecting means for projecting the coordinate values of the plurality of candidate points onto a Hough space to obtain a plurality of curves on the Hough space;
Detection interval setting means for setting a plurality of detection intervals in which a plurality of small intervals obtained by dividing the Hough space by a predetermined size are set in the Hough space;
Counting means for counting the number of the curves passing through each detection section;
Straight line calculating means for obtaining a straight line in the real space using the coordinate value of any small section in the detection section in which the number of curves passing through the detection section exceeds a first threshold value. It is a feature.

The straight line detection method of the present invention is
An image data input step for inputting image data;
A candidate point detecting step of detecting a plurality of candidate points on a straight line in real space from the input image data;
Projecting the coordinate values of the plurality of candidate points onto a Hough space to obtain a plurality of curves on the Hough space;
A detection interval setting step for setting a plurality of detection intervals in which a plurality of small intervals obtained by dividing the Hough space by a predetermined size are set in the Hough space;
A counting step for counting the number of the curves passing through each detection interval;
A straight line calculating step of obtaining a straight line in the real space using the coordinate value of any small section in the detection section in which the number of curves passing through the detection section exceeds a first threshold value. It is a feature.

The program of the present invention is a computer,
Image data input means for inputting image data;
Candidate point detecting means for detecting a plurality of candidate points on a straight line in real space from the input image data;
Projecting means for projecting the coordinate values of the plurality of candidate points onto a Hough space to obtain a plurality of curves on the Hough space;
Detection interval setting means for setting a plurality of detection intervals in which a plurality of small intervals obtained by dividing the Hough space by a predetermined size are set in the Hough space;
Counting means for counting the number of the curves passing through each detection section;
And functioning as a straight line calculating means for obtaining a straight line in the real space using the coordinate value of any small section in the detection section in which the number of curves passing through the detection section exceeds a first threshold. It is what.

  “Candidate points” refer to points that may be aligned on a straight line, and include points that are not actually aligned on a straight line.

  The “small section” is a section obtained by dividing the Hough space, and is a minimum unit for calculating the coordinate value of the intersection where a plurality of curves on the Hough space intersect.

  In addition, the straight line detection means has a number of curves that pass through a small section existing in the center of the detection section among the small sections in the detection section that exceeds the first threshold exceeds the second threshold. Sometimes, a straight line in the real space may be obtained using the coordinate value of the small section existing at the center of the detection section.

  Further, the straight line detecting means may calculate the coordinate value of the small section in which the number of curves passing through the small section exceeds the third threshold among the small sections in the detection section exceeding the first threshold. It may be used to obtain a straight line in the real space.

  According to the present invention, when determining whether or not points such as edges obtained from image data are aligned on a straight line, a wide range of detection sections are set on the Hough space for detecting a straight line. It is possible to detect a group of lines with slightly different slopes and coordinate values in the space, and obtain a straight line most suitable for representing the points that are scattered and arranged in a straight line.

  If a curve obtained by projecting points in the real space onto the Hough space obtains a straight line from the coordinate values passing through the center of a wide range of detection intervals set in the Hough space, almost all of the scattered points A straight line passing through the middle can be obtained.

  If a curve obtained by projecting points in real space onto the Hough space finds a coordinate value line that passes through a wide range of detection intervals set in the Hough space, the number of points aligned on the straight line Many straight lines can be obtained.

  Hereinafter, a specific embodiment of a straight line detection device for detecting a straight line constituting an irradiation field recognition side of a radiographic image according to the present invention will be described with reference to the drawings. FIG. 1 shows a configuration of an embodiment of a straight line detection apparatus 1 according to the present invention, and FIG. 2 is a diagram showing a radiographic image capturing apparatus using an irradiation field stop (see FIG. 2A) and the irradiation field stop. It is a figure (refer FIG.2 (b)) which shows the stimulable fluorescent substance sheet in which the irradiation field outline corresponding to an opening outline was formed.

  When performing X-ray imaging using an irradiation field stop with an imaging apparatus, first, an image obtained by imaging a subject is arranged so as to be at the center of a radiographic image (image data) P, as shown in FIG. In addition, imaging is performed with an irradiation field stop having an opening such as a rectangle between the X-ray source and the subject. The portion outside the opening is a lead plate that prevents the X-rays from reaching the subject and the stimulable phosphor sheet. In this state, when the subject is irradiated with X-rays from the X-ray source, on the radiographic image obtained from the stimulable phosphor sheet, as shown in FIG. The region corresponding to the outside (irradiation field region) Pout is not irradiated with X-rays, while the region corresponding to the inner side (irradiation field region) Pin from the opening of the irradiation field stop is directly X-rayed with the portion that has passed through the subject. The part irradiated with the line is recorded. And the part corresponding to the opening part of an irradiation field stop becomes a shape substantially the same as an opening part, and the irradiation field outside area | region Pout becomes a bright area with a high brightness | luminance. At the boundary between the irradiation field region Pin and the irradiation field outside region Pout, an irradiation field contour PS composed of a plurality of edge lines whose density changes abruptly is formed. The opening of the irradiation field stop is usually rectangular, and the irradiation field contour PS is composed of four straight irradiation field sides.

  The straight line detection device 1 of the present invention includes an image data input means 10 for inputting image data, a candidate point detection means 20 for detecting candidate points on a straight line in the real space from the input image data, and the coordinate values of the candidate points. Projection means 30 for projecting into a Hough space to obtain a curve in the Hough space, and a plurality of detection sections in the Hough space, in which a plurality of small sections obtained by dividing the Hough space by a predetermined size are set. Detection section setting means 40 for counting, counting means 50 for counting the number of curves passing through each detection section, and any of the detection sections within the detection section in which the number of curves passing through the detection section exceeds the first threshold Straight line calculation means 60 for obtaining a straight line in the real space using the coordinate values of the small section.

  Here, the operation of the straight line detection apparatus 1 will be described with reference to the flowchart of FIG.

  First, image data P of a radiographic image using an irradiation field stop obtained from the imaging apparatus is input from the image data input means 10 (S100).

  Therefore, the candidate point detection means 20 sets the center point K of the image represented by the input image data P as shown in FIG. 4A, and the end of the image data P from the center point K. Set 240 radial straight lines (α = 1.5 degrees) facing each. The set number and interval of the radial lines, the set position of the center of the radial line group, and the like can be changed as appropriate.

  Image data between pixels adjacent to each other in the direction along each of these 240 radial straight lines is compared, and two pixels having the largest difference are searched on the line. There is a large difference in pixel value between the two searched pixels, and it is a candidate point that is highly likely to be a point on the edge. In this way, candidate points E searched for in each straight line direction are detected, and a total of 240 candidate points E are obtained (see FIG. 4B, S101).

The projection unit 30 obtains a curve obtained by projecting the candidate point E on the edge in the real space (xy space) detected by the candidate point detection unit 20 onto the Hough space (ρθ space) (S102). When the coordinates of each candidate point Ei in the xy coordinate system shown in FIG. 4 (c) is (xi, yi) (i = 1, 2,..., 240), these xi and yi are used as constants in the following formula ( A curve LSi represented by 1) is obtained for each candidate point.
ρ = xi cos θ + y i sin θ (1)
As shown in FIG. 4 (c), the equation (1) represents a straight line equation passing through the coordinates (xi, yi) with the center coordinates fixed at (xi, yi), and ρ represents the coordinates (xi). , Y i), the distance between each straight line LL passing through the origin O of the xy coordinate system and θ represents the angle between the perpendicular line connecting the straight line LL and the origin O of the xy coordinate system and the x axis.

Here, each straight line LL passing through the coordinates (xi, yi) is obtained by gradually changing ρ and θ in the equation (1). In the θρ space (Hough space), the coordinates (xi, yi) are changed. These passing straight lines LL are expressed as a single curve shown in the following formula (2) (see FIG. 4C).
ρj = xcos θj + ysin θj (2)
When the same operation is performed on each edge candidate point, 240 curves LSi (i = 1, 2,..., 240) are represented in the Hough space.

  Accordingly, each intersection position (ρj, θj) where the 240 curves LSi intersect is obtained, and a straight line in the real space is obtained from the point where the intersections α, β of the plurality of curves LSi overlap each other. As shown in FIG. 5, the Hough space (ρθ space) is divided into small sections ΔD divided by Δρ and Δθ, and a detection section M in which a plurality of small sections ΔD are combined is set (S103). This is a section for determining whether or not a plurality of curves LSi subjected to Hough transform in units of small sections ΔD intersect. The sizes of Δρ and Δθ in these small sections are appropriately determined according to the resolution of the image to be detected. It is determined so that points arranged in a straight line in the space can be detected. However, the straight line constituting the irradiation field side is actually a candidate that is detected as an edge by being influenced by the intensity of the irradiated X-ray or the structure of the subject imaged at the boundary of the irradiation field side. The dots are often arranged on a straight line having a certain width. Therefore, when the number of intersections (number of votes) of the plurality of curves LSi intersects in the small section ΔD, a large number of small sections ΔD with a high vote count appear near the predetermined small section ΔD. Therefore, as shown in FIG. 5, a detection section M is set in which the small sections ΔD are grouped using a 5 × 5 mask around each small section ΔD. In each detection section, a 5 × 5 mask may be set centering on each of all the small sections ΔD. However, as shown in FIG. 6, every few small sections ΔD (× marks) are centered. You may make it set. Or each detection area M can also be made not to overlap.

  By the way, the determination as to whether or not the intersections at which a plurality of curves LSi intersect on the computer coincides is specifically performed by rounding the numerical value obtained on the computer with a predetermined digit, but the small section ΔD is It can be said that it is a section that matches the digit to be rounded.

  Specifically, the detection section M using a 5 × 5 mask may be 10 mm or less on the ρ axis and 10 ° or less on the θ axis, but (4.5 mm on the ρ axis) × (θ axis A rectangular area of about 0.8 to 1.7 ° is desirable.

  The counting means 50 obtains each intersection position (ρj, θj) where the 240 curves LSi intersect, and calculates the number (voting number) of the curves LSi intersecting at each intersection position (ρj, θj) for each detection section M. Counting is performed (S104), and a detection section M in which the number passing through each detection section M exceeds the first threshold is extracted (S105). Each detection section M corresponds to a straight line having a certain width in real space (that is, a straight line having a slightly different direction and position).

  Therefore, a small section representing a straight line that is predicted to most closely match the straight line in the real space is selected from the detection section M (S106). If the straight line candidate points in the real space vary to some extent, as shown in FIG. 7, the small section ΔD (in the circles) where many curves LSi intersect in the Hough space partially Many appear together. The range in which such a small section ΔD through which many curves LSi pass corresponds to the variation of candidate points in the real space. Therefore, the small section ΔD corresponding to the straight line in the real space is detected from the small sections ΔD included in the detection section M.

  It can be estimated that as the number of curves LSi passing through the small section ΔD increases, more candidate points are arranged in a straight line in the real space. In addition, the small section ΔD (small section indicated by the arrow in FIG. 8) existing at the center in the detection section M exceeding the first threshold is considered to be the center of variation. Therefore, when the number of curves LSi intersecting in the small section ΔD existing in the center in the detection section M exceeds the second threshold (<first threshold), the coordinate value of the small section ΔD existing in the center To find a straight line in real space.

  Alternatively, it can be estimated that among the small sections ΔD in the detection section M exceeding the first threshold value, the more intersecting curves LSi are, the more candidate points are arranged in a straight line in the real space. Therefore, the number of curves LSi intersecting in each small section ΔD is counted, and the coordinate values of the small sections where the number of curves LSi intersecting in each small section ΔD exceeds the third threshold (<first threshold) are obtained. It may be used to obtain a straight line in real space.

  Also, as shown in FIG. 9, when the straight line candidate points in the real space vary with a certain width, a large number of small sections ΔD where many curves LSi intersect in the Hough space partially appear. . Therefore, when the threshold value is lowered slightly, small sections ΔD in which the number of curves LSiLS intersecting in each small section exceed the threshold value (fourth threshold value) appear in succession. Therefore, as shown in FIG. 9, a labeling process is performed on the small section ΔD exceeding the fourth threshold value, and the small section group exceeding the fourth threshold value (# 1,. # 2, # 3, # 4, # 5) are extracted together, and the number of intersecting curves LSi is the largest from the small section group (# 1, # 2, # 3, # 4, # 5) The small section ΔD is selected, and a straight line in the real space is obtained using the coordinate value of the selected small section ΔD.

  Further, in order to compare straight lines close to each other in the Hough space, as shown in FIG. 10, the small section ΔD is set to 7 × for the small section ΔD in which the number of intersecting curves LSi exceeds the fourth threshold. Using the 15 masks, a straight line in the real space may be obtained by using the small section ΔD where the curves LSi intersect most frequently from the small sections ΔD existing in the section N collected by this mask. . When the variation of the candidate point E on the straight line is large, a straight line having an intermediate slope between the two straight lines representing two different irradiation field sides may be detected without partially appearing together. Therefore, in order to avoid this, by dividing into sections N, it is possible to separately detect a group of small sections that appear connected in the case of labeling processing.

  Specifically, the section N obtained by combining the small sections ΔD using the 7 × 15 mask may be 10 mm or less on the ρ axis and 10 ° or less on the θ axis (6.3 mm on the ρ axis). A rectangular region of about × (2.5 to 5 ° on the θ axis) is desirable.

  As described above in detail, in order to obtain a straight line representing the irradiation field side from the candidate points on the straight line that appears to some extent in the real space, such as the irradiation field side that is the boundary of the irradiation field region, By detecting a curve passing through a region corresponding to a straight line having a certain width and then selecting a straight line that is highly likely to be an irradiation field side, an accurate irradiation field side can be detected.

  Moreover, it can be made to operate | move as an irradiation field determination apparatus by installing in a computer using the medium which recorded the program which functions each said means on a computer. This program can also be provided via a network.

Schematic configuration diagram of straight line detection device The figure for explaining the method of photographing with the irradiation field stop Flow chart for explaining the operation of the straight line detection device Diagram for explaining the straight line detection method A diagram for explaining small sections and detection sections in the Hough space The figure for explaining the interval which sets the detection section A diagram showing the distribution of small sections where many curves intersect in the Hough space A diagram showing small sections and detection sections where many curves in the Hough space intersect Figure for explaining a method of extracting small sections that intersect with many nearby curves in the Hough space (part 1) Figure for explaining how to extract small sections that intersect many curves close to each other in the Hough space (Part 2)

Explanation of symbols

1 Straight line detector
10 Image data input means
20 Candidate point detection means
30 Projection means
40 Detection section setting method
50 counting means
60 Straight line calculation means P Image data

Claims (5)

Image data input means for inputting image data;
Candidate point detecting means for detecting a plurality of candidate points on a straight line in real space from the input image data;
Projecting means for projecting the coordinate values of the plurality of candidate points onto a Hough space to obtain a plurality of curves on the Hough space;
Detection interval setting means for setting a plurality of detection intervals in which a plurality of small intervals obtained by dividing the Hough space by a predetermined size are set in the Hough space;
Counting means for counting the number of the curves passing through each detection section;
Straight line calculating means for obtaining a straight line in the real space using the coordinate value of any small section in the detection section in which the number of curves passing through the detection section exceeds a first threshold value. A characteristic straight line detection device.   When the number of curves passing through the small section existing in the center of the detection section among the small sections in the detection section exceeding the first threshold exceeds the second threshold. 2. The straight line detection device according to claim 1, wherein a straight line in the real space is obtained by using a coordinate value of a small section existing at the center of the detection section.   The straight line detecting means uses the coordinate value of the small section in which the number of curves passing through the small section exceeds the third threshold among the small sections in the detection section that exceeds the first threshold. 2. The straight line detection device according to claim 1, wherein a straight line in space is obtained. An image data input step for inputting image data;
A candidate point detecting step of detecting a plurality of candidate points on a straight line in real space from the input image data;
Projecting the coordinate values of the plurality of candidate points onto a Hough space to obtain a plurality of curves on the Hough space;
A detection interval setting step for setting a plurality of detection intervals in which a plurality of small intervals obtained by dividing the Hough space by a predetermined size are set in the Hough space;
A counting step for counting the number of the curves passing through each detection interval;
A straight line calculating step of obtaining a straight line in the real space using the coordinate value of any small section in the detection section in which the number of curves passing through the detection section exceeds a first threshold value. Characteristic straight line detection method. Computer
Image data input means for inputting image data;
Candidate point detecting means for detecting a plurality of candidate points on a straight line in real space from the input image data;
Projecting means for projecting the coordinate values of the plurality of candidate points onto a Hough space to obtain a plurality of curves on the Hough space;
Detection interval setting means for setting a plurality of detection intervals in which a plurality of small intervals obtained by dividing the Hough space by a predetermined size are set in the Hough space;
Counting means for counting the number of the curves passing through each detection section;
A program that functions as a straight line calculation unit that obtains a straight line in the real space using the coordinate value of any small section in the detection section in which the number of curves passing through the detection section exceeds a first threshold.

Images