Non-contact cervical fluorescence image detection method

Abstract

The present invention proposes a non-contact cervical fluorescence image detection method, which relates to the technical field of cervical detection equipment, and includes the following steps: S1: configuring a fluorescence camera device for outputting continuous white light or fluorescence; S2: adjusting the framing part of the fluorescence camera device so that the center of the cervix is ​​located at the center position of a first reflection image; S3: the fluorescence camera device emits continuous white light to the surface of the cervix, and obtains a first reflection image of the cervix under the continuous white light; preprocessing the first reflection image to obtain the boundary range of the cervix; extracting feature points within the boundary range of the cervix; S4: maintaining the position of the framing part of the fluorescence camera device unchanged, allowing the fluorescence camera device to emit fluorescence to the surface of the cervix, and obtaining a second reflection image under fluorescence; S5: extracting image information of each pixel point based on the obtained second reflection image, judging whether there is a fluorescence point in the second reflection image, and converting the fluorescence point into the first reflection image.

Description

Technical Field

[0001] The present invention relates to the technical field of cervical detection equipment, and in particular to a non-contact cervical fluorescence image detection method. Background Art

[0002] Cervical cancer is a cancer that occurs in the female cervix. It ranks second among the most common malignant tumors in women. There are more than 500,000 new confirmed cases worldwide each year, of which about 140,000 women in my country are diagnosed with cervical cancer each year. With the change of lifestyle, the age of patients is showing a trend of younger age. It has been confirmed that long-term repeated infection with human papillomavirus (HPV) is the main cause of cervical cancer. From HPV infection to the development of cervical cancer, the condition can develop into three stages: CINI, CIN II and CIN III, until it finally develops into invasive cancer. The earlier the diagnosis of cervical cancer is, the more beneficial it is for the treatment of the disease, which can significantly reduce the incidence and mortality of cervical cancer. The commonly used inspection methods are naked eye observation, glacial acetic acid coating observation, conventional colposcopy combined with acetic acid and Lugol's iodine test, Pap smear, liquid-based cytology, etc. However, the above methods have a high false negative rate, poor sensitivity, and require a long observation time. The cervix needs to be contact coated, which may cause discomfort to the patient, and the contact process may introduce other pathogens. None of the above methods can accurately determine the nature of the lesion, and it is often necessary to combine cervical biopsy pathological examination for diagnosis. The biopsy often takes more than a week, which brings great mental burden and pressure to the patient.

[0003] Therefore, it is very necessary to provide a non-contact, cervical fluorescence image detection method that does not require the application of reagents on the cervix and shortens the waiting time for detection. Summary of the invention

[0004] In view of this, the present invention proposes a non-contact cervical fluorescence image detection method without human body stimulation and with a shorter detection process.

[0005] The technical solution of the present invention is implemented as follows: The present invention provides a non-contact cervical fluorescence image detection method, comprising the following steps:

[0006] S1: A fluorescence camera is configured to output continuous white light or fluorescence;

[0007] S2: using a dilator to expose the cervix, and adjusting the viewfinder of the fluorescent camera device so that the center of the cervix is ​​located at the center of the first reflected image;

[0008] S3: The fluorescent camera emits continuous white light to the surface of the cervix and obtains a first reflection image of the cervix under the continuous white light; preprocesses the first reflection image to obtain the boundary range of the cervix; extracts feature points within the boundary range of the cervix;

[0009] S4: maintaining the position of the viewfinder of the fluorescent camera device unchanged, allowing the fluorescent camera device to emit fluorescence to the surface of the cervix, and obtaining a second reflection image under the fluorescence;

[0010] S5: extracting image information of each pixel according to the acquired second reflected image, determining whether there is a fluorescent point in the second reflected image, and converting the fluorescent point into the first reflected image.

[0011] On the basis of the above technical scheme, preferably, the fluorescent photography device includes a continuous white light source, a fluorescent light source, a beam combiner, a lens, a binocular depth camera and a CCD, the output end of the continuous white light source and the output end of the fluorescent light source are respectively connected to the first input end and the second input end of the beam combiner in a one-to-one optical path, and the output end of the beam combiner points in the direction of the cervix; the lens is arranged facing the cervix, and a CCD or a binocular depth camera is arranged on the side of the lens away from the cervix; the optical axis of the lens selectively faces the optical axis of the CCD or the binocular depth camera; the continuous white light source or the fluorescent light source is selectively output to the beam combiner, and the CCD is used to obtain the second reflected image; the binocular depth camera is used to obtain the depth image and the first reflected image of the cervical surface.

[0012] Preferably, the framing part of the fluorescent camera device is adjusted so that the center of the cervix is ​​located at the center of the first reflected image. An image coordinate system is constructed with the center of the image acquired by the left camera of the binocular depth camera as the origin, and the width direction of the image coordinate system is the x-axis of the image coordinate system, and the height direction of the image is the y-axis of the image coordinate system. A continuous external cervical os area is found within the neighborhood of the origin of the image coordinate system. The center of the first reflected image acquired by the left camera of the binocular depth camera is used as the starting point to draw a number of independent first line segments, wherein each independent first line segment is set at an angle, and the length of each independent first line segment does not exceed half of the length or width of the first reflected image. A first mutation point of the RGB accumulated value of the pixel point on each independent first line segment is found, and a closed curve is fitted according to each first mutation point, and the center point of the fitted closed curve is found, and the center point of the closed curve is coincident with the center position of the first reflected image; the optical axis of the CCD is made to coincide with the optical axis of the left camera of the binocular depth camera, and the posture of the CCD and the left camera of the binocular depth camera is calibrated.

[0013] Preferably, the first mutation point of the RGB cumulative value of the pixel points on each independent first line segment refers to the RGB cumulative value of a pixel point on the independent first line segment having a mutation of more than 35% with the RGB cumulative value of the adjacent pixel point, and the pixel point with the smaller RGB cumulative value is retained as the first mutation point; all the first mutation points on each independent first line segment are obtained, the first mutation points are fitted into a rectangle or an ellipse, and the center of the rectangular or elliptical external cervical os area is translated to the center position of the first reflected image.

[0014] Further preferably, the preprocessing of the first reflected image in step S3 to obtain the boundary range of the cervix is ​​to find a continuous cervical external os area, and after fitting a closed curve, generate a filling pattern covering the fitted closed curve according to the average value of the RGB of the adjacent pixels outside the fitted closed curve; take the center of the first reflected image as the starting point, draw a number of rays, and find on the rays that satisfy the inequality At least one pixel point in the image is taken as a boundary point, where R, G and B are respectively the normalized values ​​of the color channel values, and the value range is (0, 1); the boundary points on each ray are obtained and fitted into the cervical contour. If the cervical contour displayed on the first reflected image is incomplete, the first reflected image is rotated, translated or scaled based on the center position so that the cervical contour is completely displayed, and the length extension direction of the cervical contour is parallel to the horizontal axis of the image coordinate system, and the width extension direction of the cervical contour is parallel to the vertical axis of the image coordinate system; then the content of the first reflected image outside the cervical contour is cropped, the filling pattern is removed, and a number of feature points are extracted from the cervical contour range.

[0015] More preferably, a number of feature points are extracted from the cervical contour range, and a world coordinate system is constructed with the center of the left camera of the binocular depth camera, and the center line of the left camera and the right camera of the binocular depth camera is connected to the X coordinate system of the world coordinate system. C Axis parallel, Y of the world coordinate system C The Z axis is parallel to the y axis of the image coordinate system of the left camera of the stereo depth camera, and the optical axis of the left camera of the stereo depth camera is used as the Z axis. C Axis; let the coordinates of the point on the cervical surface in the world coordinate system be P(X C , Y C , Z C ), the coordinates of the corresponding points of the two depth images obtained by the binocular depth camera are P L (X L , Y L ) and P R (X R , Y R ), the optical axes of the left and right cameras of the binocular depth camera are both located at X C Z C plane, and the distance is b, the focal length of the left camera is f, then Then the depth information of the point on the cervical surface in the world coordinate system is The parameter d is the disparity of the image of the left camera and the image of the right camera of the binocular depth camera for any point on the left side of the world; the corresponding depth information D is assigned to each pixel point within the cervical contour range of the image coordinate system of the left camera of the depth camera; a plurality of second line segments are constructed, one end of the second line segment is located at the boundary of the external cervical os area, and the other end of the second line segment is located on the cervical contour, and the pixel point where the depth information change rate suddenly changes on the second line segment is found as the first feature point, the second feature point is extracted from the cervical contour, and the first feature point and the second feature point are marked and retained respectively.

[0016] Further preferably, the pixel point where the depth information change rate suddenly changes on the second line segment is found as the first feature point, the second feature point is extracted from the cervical contour, and the first feature point and the second feature point are marked and retained respectively, by performing the following steps: 1) adjacent second line segments are set at a certain angle or spacing interval; 2) with the x-axis of the image coordinate system as the horizontal axis and the depth information as the vertical axis, a quadratic curve is fitted to the depth information of each pixel point on the current second line segment, and a pixel point located outside the curve and at a distance from the curve exceeding 3 times the diagonal length of the pixel point is found, and the pixel point is marked and retained as the first feature point on the current second line segment; 3) if the first feature point exists on the current second line segment, a number of parallel line segments are drawn symmetrically on both sides of the current second line segment with an offset distance of one pixel diagonal length, and the first feature point is found on other parallel line segments within the neighborhood of the current second line segment according to the content of step 2), until all parallel line segments are searched, and the number of parallel line segments on both sides of the current second line segment is D ist is the maximum distance from the first feature point on the current second line segment to the fitted quadratic curve, L pixel is the diagonal length of a single pixel, is the rounding operation, n is the correction term, if there is more than one first feature point on the current second line segment and they are set adjacently, the value of n is [1, 2], and the value of n is determined by the number of adjacent first feature points; if there is more than one first feature point on the current second line segment and they are not set adjacently, the value of n is [3, 5], and the value of n is determined by the spacing between non-adjacent first feature points and D ist ; 5) If the first feature point does not exist on the current second line segment, the current second line segment is rotated by a certain angle along the center of the image coordinate system to construct a new second line segment, and the above steps 2)-4) are repeated; 6) The farthest pixel point of the cervical contour along the x-axis or y-axis of the image coordinate system is obtained, and it is marked and retained as the second feature point.

[0017] Preferably, the specific content of step S4 is to move the binocular depth camera horizontally along the ground, move the CCD along the optical axis, so that the light emitted by the continuous white light source can be clearly imaged after being reflected by the surface of the cervix and entering the CCD, and then switch the white light source to a fluorescent light source, the fluorescent light source contains at least one specific wavelength, and the CCD obtains fluorescent images of corresponding wavelengths respectively.

[0018] More preferably, in order to enable the light emitted by the continuous white light source to be clearly imaged after being reflected from the surface of the cervix and entering the CCD, a fluorescent image coordinate system based on the CCD is constructed, wherein the horizontal axis of the fluorescent image coordinate system is parallel to the x-axis of the image coordinate system, and the vertical axis of the fluorescent image coordinate system is parallel to the y-axis of the image coordinate system; in the cervical surface image acquired by the CCD under the continuous white light source, the cervical contour is completely displayed, and the farthest point of the cervical contour from the far point of the fluorescent image coordinate system is located on the horizontal axis or the vertical axis of the fluorescent image coordinate system based on the CCD, and after switching to the fluorescent light source, bright spots in the grayscale fluorescent image at different specific wavelengths are acquired.

[0019] More preferably, the fluorescent light source comprises at least one specific wavelength, which refers to a combination of one or more of 470 nm, 498 nm, 675 nm or 905 nm.

[0020] Compared with the prior art, the non-contact cervical fluorescence image detection method provided by the present invention has the following beneficial effects:

[0021] (1) The present application configures a fluorescent camera to acquire images under continuous white light and fluorescent light in stages, aligns, fits boundaries or cuts the reflection images acquired under different light sources, and fuses the feature points under different light sources into one image, so as to facilitate the identification and statistics of the range of removed pixels under different light sources, and provide a reliable image basis for contactless diagnosis;

[0022] (2) The present application does not adopt the conventional glacial acetic acid coating and staining observation method, but instead adopts a combination of white light and fluorescence to scan the same area multiple times and mark the pixels of interest by superimposing the images. There is no foreign body sensation or discomfort, and the inspection time can be greatly shortened. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0024] Figure 1It is a schematic diagram of the process of the non-contact cervical fluorescence image detection method of the present invention;

[0025] Figure 2 The present invention is a schematic structural diagram of a fluorescence photography device for the non-contact cervical fluorescence image detection method of the present invention. DETAILED DESCRIPTION

[0026] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0027] like Figure 1 As shown, the present invention provides a non-contact cervical fluorescence image detection method, comprising the following steps:

[0028] S1: Configure a fluorescent camera device to output continuous white light or fluorescent light.

[0029] like Figure 2 As shown, the fluorescent photography device used in the present application includes a continuous white light source 11, a fluorescent light source 12, a beam combiner 13, a lens 14, a binocular depth camera 15 and a CCD 16. The output end of the continuous white light source 11 and the output end of the fluorescent light source 12 are respectively connected to the first input end and the second input end of the beam combiner 13 in a one-to-one optical path, and the output end of the beam combiner 13 points to the direction of the cervix; the lens 14 is arranged directly opposite to the cervix, and a CCD 16 or a binocular depth camera 15 is arranged on the side of the lens 14 away from the cervix; the optical axis of the lens 14 selectively faces the optical axis of the CCD 16 or the binocular depth camera 15; the continuous white light source 11 or the fluorescent light source 12 selectively outputs to the beam combiner 13, and the CCD 16 is used to obtain the second reflected image; the binocular depth camera 15 is used to obtain the depth image and the first reflected image of the cervical surface. The beam combiner 13 can use a Y-shaped optical fiber. The binocular depth camera 15 can be implemented using Knicet2.0 or similar products. CCD 16 can image both continuous white light source 11 and fluorescent light source 12. Lens 14 can focus light and better focus on the local area of ​​cervix.

[0030] S2: Use a dilator to dilate the vagina to expose the cervix, and adjust the viewfinder of the fluorescent camera device so that the center of the cervix is ​​located at the center of the first reflected image.

[0031] Specifically, the width direction of the image coordinate system is used as the x-axis of the image coordinate system, and the height direction of the image is used as the y-axis of the image coordinate system. Within the neighborhood of the origin of the image coordinate system, a continuous cervical external os area is searched, and the center of the first reflection image obtained by the left camera of the binocular depth camera 15 is used as the starting point to draw a number of independent first line segments, each of which is set at an angle, and the length of each independent first line segment does not exceed half of the length or width of the first reflection image. The first mutation point of the RGB cumulative value of the pixel point on each independent first line segment is found, and a closed curve is fitted according to each first mutation point, and the center point of the fitted closed curve is found, and the center point of the closed curve is overlapped with the center position of the first reflection image; the optical axis of CCD16 is overlapped with the optical axis of the left camera of the binocular depth camera 15, and the posture of CCD16 and the left camera of the binocular depth camera 15 is calibrated. In this application, the left camera of the binocular depth camera 15 is the main camera, and the right camera is the auxiliary calibration camera. Therefore, the image coordinate system is established with the left camera.

[0032] In the above content, the first mutation point refers to the RGB cumulative value of a pixel point on the independent first line segment and the RGB cumulative value of the adjacent pixel point, which has a mutation of more than 35%, and the pixel point with a smaller RGB cumulative value is retained as the first mutation point; all the first mutation points on each independent first line segment are obtained, and the first mutation points are fitted into a rectangle or an ellipse, and the center of the rectangular or elliptical cervical external os area is translated to the center position of the first reflected image. The first mutation point indicates the boundary of the dark area near the cervical external os, which is used to define the boundary of the inner edge of the cervical image.

[0033] S3: The fluorescent camera device emits continuous white light to the surface of the cervix and obtains a first reflection image of the cervix under the continuous white light; preprocesses the first reflection image to obtain the boundary range of the cervix; and extracts feature points within the boundary range of the cervix.

[0034] In this step, the first reflected image is preprocessed to obtain the boundary range of the cervix, to find a continuous cervical external os area, and after fitting a closed curve, a filling pattern covering the fitted closed curve is generated according to the average value of the RGB of the adjacent pixels outside the fitted closed curve; starting from the center of the first reflected image, several rays are drawn, and the rays are searched for the inequality satisfying the inequality At least one pixel point in is taken as a boundary point, where R, G and B are respectively the normalized values ​​of the color channel values, and the value range is (0, 1]; normalization is to perform the operation of [current color channel value - current color channel minimum value] / [current color channel maximum value - color channel minimum value], and the value range of the color channel is [0, 255], so a decimal within (0, 1] can be obtained, and the position of the dark edge outside the cervical contour is determined according to the above inequality, thereby determining one or more boundary points of the cervical contour.

[0035] Then, the boundary points on each ray are obtained and fitted into the cervical contour. If the cervical contour displayed on the first reflected image is incomplete, the first reflected image is rotated, translated or scaled based on the center position so that the cervical contour is fully displayed, and the length extension direction of the cervical contour is parallel to the horizontal axis of the image coordinate system, and the width extension direction of the cervical contour is parallel to the vertical axis of the image coordinate system; then the content of the first reflected image outside the cervical contour is cropped, and the filling pattern is removed. The area inside the cervical contour and outside the external cervical os together constitute the image of the cervix. Then, several feature points are extracted from the range of the cervical contour.

[0036] Since the binocular depth camera 15 and CCD16 are coaxial and block each other, when the binocular depth camera 15 is removed and CCD16 is used, although the usage time is very short, there may still be changes in the patient's posture. In order to facilitate subsequent image fusion or comparison, some feature points need to be selected to define the cervical contour in the image coordinate system, as well as the cervical surface defects observable by the binocular depth camera 15, to facilitate further analysis.

[0037] A number of feature points are extracted from the cervical contour range, and a world coordinate system is constructed with the center of the left camera of the binocular depth camera 15, and the center line of the left camera and the right camera of the binocular depth camera 15 and the X coordinate of the world coordinate system are connected. C Axis parallel, Y of the world coordinate system C The Z axis is parallel to the y axis of the image coordinate system of the left camera of the binocular depth camera 15, and the optical axis of the left camera of the binocular depth camera 15 is used as the Z axis. C Axis; let the coordinates of the point on the cervical surface in the world coordinate system be P(X C , Y C , Z C ), the coordinates of the corresponding points of the two depth images acquired by the binocular depth camera 15 are P L (X L , Y L ) and P R (X R , Y R ), the optical axes of the left camera and the right camera of the binocular depth camera 15 are both located at X C Z Cplane, and the distance is b, the focal length of the left camera is f, then Then the depth information of the point on the cervical surface in the world coordinate system is Parameter d is the parallax of any point on the left side of the world in the image of the left camera and the image of the right camera of the binocular depth camera 15; assign corresponding depth information D to each pixel point within the cervical contour range of the image coordinate system of the left camera of the depth camera; construct a number of second line segments, one end of the second line segment is located at the boundary of the external cervical os area, and the other end of the second line segment is located on the cervical contour, find the pixel point on the second line segment where the depth information change rate suddenly changes as the first feature point, extract the second feature point from the cervical contour, and mark and retain the first feature point and the second feature point respectively. The first feature point corresponds to the cervical surface defect observable by the binocular depth camera 15, such as abnormal protrusions or depressions on the oval cervical surface;

[0038] Specifically, the pixel point where the depth information change rate suddenly changes on the second line segment is found as the first feature point, the second feature point is extracted from the cervical contour, and the first feature point and the second feature point are marked and retained respectively, and the following steps are performed: 1) adjacent second line segments are set at a certain angle or spacing interval; 2) with the x-axis of the image coordinate system as the horizontal axis and the depth information as the vertical axis, a quadratic curve is fitted to the depth information of each pixel point on the current second line segment, and a pixel point located outside the curve and with a distance from the curve exceeding 3 times the diagonal length of the pixel point is found, and the pixel point is marked and retained as the first feature point on the current second line segment; 3) if the first feature point exists on the current second line segment, a number of parallel line segments are drawn symmetrically on both sides of the current second line segment with an offset distance of one pixel diagonal length, and the first feature point is found on other parallel line segments within the neighborhood of the current second line segment according to the content of step 2), until all parallel line segments are searched, and the number of parallel line segments on both sides of the current second line segment is D ist is the maximum distance from the first feature point on the current second line segment to the fitted quadratic curve, L pixel is the diagonal length of a single pixel, is the rounding operation, n is the correction term, if there is more than one first feature point on the current second line segment and they are set adjacently, the value of n is [1, 2], and the value of n is determined by the number of adjacent first feature points; if there is more than one first feature point on the current second line segment and they are not set adjacently, the value of n is [3, 5], and the value of n is determined by the spacing between non-adjacent first feature points and D ist; 5) If the first feature point does not exist on the current second line segment, the current second line segment is rotated by a certain angle along the center of the image coordinate system to construct a new second line segment, and the above steps 2)-4) are repeated; 6) The farthest pixel point of the cervical contour along the x-axis or y-axis of the image coordinate system is obtained, and it is marked and retained as the second feature point. The second feature point indicates the characteristic position of the cervical contour.

[0039] S4: Maintaining the position of the viewfinder of the fluorescent camera device unchanged, the fluorescent camera device emits fluorescence to the surface of the cervix, and obtains a second reflected image under the fluorescence.

[0040] Preferably, the specific content of step S4 is to move the binocular depth camera 15 horizontally along the ground, and move the CCD16 along the optical axis so that the light emitted by the continuous white light source 11 can be clearly imaged after being reflected by the surface of the cervix and entering the CCD16, and then the white light source is switched to the fluorescent light source 12, the fluorescent light source 12 contains at least one specific wavelength, and the CCD16 obtains the fluorescent images of the corresponding wavelengths respectively.

[0041] More preferably, in order to enable the light emitted by the continuous white light source 11 to be clearly imaged after being reflected by the surface of the cervix and entering the CCD16, a fluorescence image coordinate system based on CCD16 is constructed, wherein the horizontal axis of the fluorescence image coordinate system is parallel to the x-axis of the image coordinate system, and the vertical axis of the fluorescence image coordinate system is parallel to the y-axis of the image coordinate system; in the cervical surface image acquired by CCD16 under the continuous white light source 11, the cervical contour is completely displayed, and the farthest point of the cervical contour from the far point of the fluorescence image coordinate system is located on the horizontal axis or the vertical axis of the fluorescence image coordinate system based on CCD16.

[0042] The fluorescent light source 12 includes at least one specific wavelength, which is a combination of at least one of 470 nm, 498 nm, 675 nm, or 905 nm. If multiple wavelengths are selected, multiple fluorescent images are generated accordingly.

[0043] If it is a discrete single fluorescent pixel that is not adjacent to any other fluorescent pixel, it can be ignored. If it is an adjacent and clustered fluorescent pixel, the convex hull of OPENCV is used to construct a curve that includes all adjacent and clustered fluorescent pixels, and the constructed curve is retained.

[0044] S5: extracting image information of each pixel according to the acquired second reflected image, determining whether there is a fluorescent point in the second reflected image, and converting the fluorescent point into the first reflected image.

[0045] After switching to the fluorescent light source 12, the bright spots in the grayscale fluorescent images at different specific wavelengths are obtained, and the bright spots are converted into the image coordinate system, and are marked and saved as the third feature points. Thus, the first reflected image based on the image coordinate system includes the first feature point, the curve that includes all adjacent and aggregated fluorescent pixels, and other features of interest, which are convenient for medical staff to observe and confirm.

[0046] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A non-contact cervical fluorescence image detection method, characterized in that: The steps include: S1: A fluorescence camera is configured to output continuous white light or fluorescence; S2: using a dilator to expose the cervix, and adjusting the viewfinder of the fluorescent camera device so that the center of the cervix is ​​located at the center of the first reflected image; S3: The fluorescent camera emits continuous white light to the surface of the cervix and obtains a first reflection image of the cervix under the continuous white light; preprocesses the first reflection image to obtain the boundary range of the cervix; extracts feature points within the boundary range of the cervix; S4: maintaining the position of the viewfinder of the fluorescent camera device unchanged, allowing the fluorescent camera device to emit fluorescence to the surface of the cervix, and obtaining a second reflection image under the fluorescence; S5: extracting image information of each pixel according to the acquired second reflected image, determining whether there is a fluorescent point in the second reflected image, and converting the fluorescent point into the first reflected image; The fluorescent photography device comprises a continuous white light source (11), a fluorescent light source (12), a beam combiner (13), a lens (14), a binocular depth camera (15) and a CCD (16); the output end of the continuous white light source (11) and the output end of the fluorescent light source (12) are respectively connected to the first input end and the second input end of the beam combiner (13) in a one-to-one optical path correspondence, and the output end of the beam combiner (13) points in the direction of the cervix; the lens (14) is arranged facing the cervix, and a CCD (16) or a binocular depth camera (15) is arranged on a side of the lens (14) away from the cervix; the optical axis of the lens (14) selectively faces the optical axis of the CCD (16) or the binocular depth camera (15); the continuous white light source (11) or the fluorescent light source (12) is selectively output to the beam combiner (13), and the CCD (16) is used to obtain a second reflected image; the binocular depth camera (15) is used to obtain a depth image and a first reflected image of the cervical surface; The framing part of the fluorescent camera device is adjusted so that the center of the cervix is ​​located at the center of the first reflected image. An image coordinate system is constructed with the center of the image acquired by the left camera of the binocular depth camera (15) as the origin, the width direction of the image coordinate system is the x-axis of the image coordinate system, and the height direction of the image is the y-axis of the image coordinate system. A continuous cervical external os area is found within the neighborhood of the origin of the image coordinate system. A plurality of independent first line segments are drawn with the center of the first reflected image acquired by the left camera of the binocular depth camera (15) as the starting point. The independent first line segments are arranged at an angle, and the length of each independent first line segment does not exceed half the length or width of the first reflected image. A first mutation point of the RGB accumulated value of the pixel point on each independent first line segment is found. A closed curve is fitted according to each first mutation point, the center point of the fitted closed curve is found, and the center point of the closed curve is made to coincide with the center position of the first reflected image. The optical axis of the CCD (16) is made to coincide with the optical axis of the left camera of the binocular depth camera (15), and the position and posture of the CCD (16) and the left camera of the binocular depth camera (15) are calibrated. The method of finding the first mutation point of the RGB cumulative value of the pixel points on each independent first line segment refers to that the RGB cumulative value of a pixel point on the independent first line segment undergoes a mutation of more than 35% with the RGB cumulative value of the adjacent pixel point, and the pixel point with the smaller RGB cumulative value is retained as the first mutation point; all the first mutation points on each independent first line segment are obtained, the first mutation points are fitted into a rectangle or an ellipse, and the center of the rectangular or elliptical cervical external os area is translated to the center position of the first reflected image.

2. The non-contact cervical fluorescence image detection method according to claim 1, characterized in that: The preprocessing of the first reflected image in step S3 to obtain the boundary range of the cervix is ​​to find a continuous cervical external os area, and after fitting a closed curve, generate a filling pattern covering the fitted closed curve according to the average value of the RGB of the adjacent pixels outside the fitted closed curve; take the center of the first reflected image as the starting point, draw several rays, and find on the rays that satisfy the inequality At least one pixel point in the image is taken as a boundary point, where R, G and B are respectively the normalized values ​​of the color channel values, and the value range is (0, 1); the boundary points on each ray are obtained and fitted into the cervical contour. If the cervical contour displayed on the first reflected image is incomplete, the first reflected image is rotated, translated or scaled based on the center position so that the cervical contour is completely displayed, and the length extension direction of the cervical contour is parallel to the horizontal axis of the image coordinate system, and the width extension direction of the cervical contour is parallel to the vertical axis of the image coordinate system; then the content of the first reflected image outside the cervical contour is cropped, the filling pattern is removed, and a number of feature points are extracted from the cervical contour range.

3. The non-contact cervical fluorescence image detection method according to claim 2, characterized in that: A number of feature points are extracted from the cervical contour range, and a world coordinate system is constructed with the center of the left camera of the binocular depth camera (15), and the center line of the left camera and the right camera of the binocular depth camera (15) is connected to the X coordinate of the world coordinate system. C Axis parallel, Y of the world coordinate system C The Z axis is parallel to the y axis of the image coordinate system of the left camera of the binocular depth camera (15), and the optical axis of the left camera of the binocular depth camera (15) is used as the Z axis. C Axis; let the coordinates of the point on the cervical surface in the world coordinate system be P(X C , Y C , Z C ), the coordinates of the corresponding points of the two depth images obtained by the binocular depth camera (15) are P L (X L , Y L ) and P R (X R , Y R ), the optical axes of the left camera and the right camera of the binocular depth camera (15) are both located at X C Z C plane, and the distance is b, the focal length of the left camera is f, then Then the depth information of the point on the cervical surface in the world coordinate system is The parameter d is the parallax of the image of the left camera and the image of the right camera of the binocular depth camera (15) at any point in the world coordinate system; the corresponding depth information D is assigned to each pixel point within the cervical contour range of the image coordinate system of the left camera of the depth camera; a plurality of second line segments are constructed, one end of the second line segment is located at the boundary of the cervical external os area, and the other end of the second line segment is located on the cervical contour, and the pixel point at which the depth information change rate suddenly changes on the second line segment is found as the first feature point, the second feature point is extracted from the cervical contour, and the first feature point and the second feature point are marked and retained respectively.

4. The non-contact cervical fluorescence image detection method according to claim 3, characterized in that: Find the pixel point where the depth information change rate on the second line segment suddenly changes as the first feature point, extract the second feature point from the cervical contour, mark and retain the first feature point and the second feature point respectively, and perform the following steps: 1) adjacent second line segments are set at a certain angle or spacing interval; 2) with the x-axis of the image coordinate system as the horizontal axis and the depth information as the vertical axis, perform quadratic curve fitting on the depth information of each pixel point on the current second line segment, find the pixel point that is outside the curve and the distance from the curve exceeds 3 times the diagonal length of the pixel point, mark the pixel point as the first feature point on the current second line segment and retain it; 3) if the first feature point exists on the current second line segment, draw several parallel line segments symmetrically on both sides of the current second line segment with one times the diagonal length of the pixel as the offset distance, and find whether the first feature point exists on other parallel line segments within the neighborhood of the current second line segment according to the content of step 2), until all parallel line segments are searched, and the number of parallel line segments on both sides of the current second line segment is D ist is the maximum distance from the first feature point on the current second line segment to the fitted quadratic curve, L pixel is the diagonal length of a single pixel, is the rounding operation, n is the correction term, if there is more than one first feature point on the current second line segment and they are set adjacently, the value of n is [1, 2], and the value of n is determined by the number of adjacent first feature points; if there is more than one first feature point on the current second line segment and they are not set adjacently, the value of n is [3, 5], and the value of n is determined by the spacing between non-adjacent first feature points and D ist ; 5) If the first feature point does not exist on the current second line segment, the current second line segment is rotated by a certain angle along the center of the image coordinate system to construct a new second line segment, and the above steps 2)-4) are repeated; 6) The farthest pixel point of the cervical contour along the x-axis or y-axis of the image coordinate system is obtained, and it is marked and retained as the second feature point.

5. The non-contact cervical fluorescence image detection method according to claim 3, characterized in that: The specific content of step S4 is to horizontally move the binocular depth camera (15) along the ground, move the CCD (16) along the optical axis, so that the light emitted by the continuous white light source (11) can be clearly imaged after being reflected by the surface of the cervix and entering the CCD (16), and then switch the white light source to a fluorescent light source (12), the fluorescent light source (12) contains at least one specific wavelength, and the CCD (16) respectively obtains fluorescent images of corresponding wavelengths.

6. The non-contact cervical fluorescence image detection method according to claim 5, characterized in that: The light emitted by the continuous white light source (11) is reflected by the surface of the cervix and enters the CCD (16) to form a clear image. A fluorescence image coordinate system based on the CCD (16) is constructed, wherein the horizontal axis of the fluorescence image coordinate system is parallel to the x-axis of the image coordinate system, and the vertical axis of the fluorescence image coordinate system is parallel to the y-axis of the image coordinate system; in the cervical surface image obtained by the CCD (16) under the continuous white light source (11), the cervical contour is completely displayed, and the farthest point of the cervical contour from the far point of the fluorescence image coordinate system is located on the horizontal axis or the vertical axis of the fluorescence image coordinate system based on the CCD (16); after switching to the fluorescence light source (12), bright spots in the grayscale fluorescence image at different specific wavelengths are obtained.

7. The non-contact cervical fluorescence image detection method according to claim 5, characterized in that: The fluorescent light source (12) includes at least one specific wavelength, which is a combination of at least one wavelength of 470 nm, 498 nm, 675 nm or 905 nm.

Images