CN110349171A - A kind of scoliosis back contour curve extracting method based on gray scale intermediate value - Google Patents

Abstract

The invention discloses a scoliosis back contour curve extraction method based on the gray scale median value. Firstly, a blue background cloth and a photographic lighting lamp are arranged, and an ordinary digital camera is used to take an image of the back of an upright human body, and then the red channel information of the RGB image is extracted. , the back contour of the human body is obtained by binarization processing, and then the back image is grayscaled and multiplied with the binarized human back contour map to obtain a grayscale image that removes background interference, and then the grayscale image is subtracted from the pixel grayscale The average value is used to obtain a new grayscale image, and then the grayscale median pixel feature points are calculated line by line, and then the feature points are subjected to polynomial fitting by the least square method to obtain the spine contour fitting curve. The method of the invention is simple and practical, and the algorithm is easy to realize, and can effectively complete the purpose of extracting the contour curve of the back spine of the human body.

Description

A scoliosis back contour curve extraction method based on gray value

technical field

The invention belongs to the technical field of image processing, and in particular relates to a scoliosis back contour curve extraction method based on gray-scale median value.

Background technique

The spine is the central axis of the human body. When scoliosis is severe, it will not only cause abnormal body appearance and motor dysfunction, but also cause cardiopulmonary dysfunction due to thoracic deformity, reduce the quality of life, and seriously affect the physical and mental health of adolescents. If the disease is not detected early and treated actively, it will not only affect the child's body shape and appearance, but also may cause abnormal cardiopulmonary function, premature spine degeneration, pain, and trunk imbalance. Sick children with severe deformities may even suffer early cardiopulmonary failure, leading to death.

There are many methods for checking scoliosis, which can be roughly divided into two categories: physical measurement and image measurement. Physical measurement refers to the measurement of scoliosis in direct contact with the back of the human body, mainly including the Adams forward bending test, the application of the scoliosis ruler to measure the rotation angle of the trunk, and the measurement of the rib protuberance; image measurement refers to methods that do not directly contact the back of the human body Contact inspection methods mainly include Moiré image measurement method, X-ray film measurement method, structured light measurement method, laser scanner measurement method, etc.

Although the existing methods can check scoliosis, because most of the existing census methods are based on manual physical detection, manual detection is cumbersome and inefficient when conducting a general survey on a large number of people, especially adolescents. Staff fatigue can also lead to errors and miscalculations. Carrying out censuses with X-ray films will cause a lot of unnecessary radiation damage to teenagers, especially children, and the cost is relatively high.

Contents of the invention

Purpose of the invention: To provide a scoliosis back contour curve extraction method based on the gray value median, which is non-destructive and radiation-free, simple and practical, fast and effective, and the algorithm is easy to implement, which can effectively extract the spine back contour curve and complete the examination of scoliosis Task.

Technical solution: In order to realize the above-mentioned invention purpose, the present invention adopts the following technical solutions:

A method for extracting scoliosis back contour curve based on gray value, comprising the following steps:

(1) Arrange blue background cloth and photographic lighting, and the lighting should be evenly irradiated on the back of the human body;

(2) Use a digital camera to capture images of the back of the human body, and collect color image information on the back of the human body upright, including RGB three-channel brightness information and image resolution size n×m;

(3) grayscale processing is carried out to the color image of the back of the human body acquired in step (2), to obtain a grayscale image;

(4) Carry out channel separation to the color image of the back of the human body acquired in step (2), and obtain the grayscale image information of the red channel;

(5) Binarize the red channel grayscale image in step (4) to obtain a binary image of the back of the human body;

(6) dot product the human body back grayscale image in step (3) and the human body back binary image in step (5), obtain the back grayscale image that removes background interference;

(7) Subtract the image grayscale base value from the back grayscale image in step (6), to obtain a new grayscale image;

(8) obtain the gray-scale median pixel feature point line by line to the new gray-scale image that step (7) obtains;

(9) Fit the feature points obtained in step (8) into a characteristic curve of the back contour of the spine.

Further, when collecting the color image of the back of the human body in step (2), the person to be collected needs to stand upright, and the image includes the uppermost end of the cervical spine to the lowermost end of the lumbar spine.

Further, the color image of the back of the human body obtained in step (2) is obtained according to the formula Gray1(i,j)=0.30R(i,j)+0.59G(i,j)+0.11B(i,j) in step (3). ) to perform RGB three-component weighted average processing to obtain grayscale image Gray1, R(i,j), G(i,j), B(i,j) are the brightness values of red, green and blue three-channel component pixels respectively, Gray1(i , j) is the gray value of the gray image pixel obtained.

Further, in step (4), the RGB three-component channel separation is performed on the human body back color image obtained in step (2), and the red channel image information is extracted separately according to the formula Gray2(i,j)=R(i,j) to obtain the red channel In the grayscale image Gray2, R(i,j) is the brightness value of the pixel of the red channel component, and Gray2(i,j) is the grayscale value of the pixel of the grayscale image of the red channel.

Further, converting the red channel grayscale image Gray2 into a binary image in step (5) specifically includes the following sub-steps:

(51) Set the gray scale threshold T;

(52) Binarize the red channel grayscale image Gray2 obtained in step (4) according to the grayscale threshold T set in (51), if Gray2(i,j)>T, then Binary(i,j) =1; if Gray2(i,j)≤T, then Binary(i,j)=0, to obtain the binary image Binary1 of the back contour of the human body.

Further, in step (6), according to the formula Gray3(i,j)=Gray1(i,j)*Binary(i,j), the grayscale image Gray1 obtained in step (3) and the human body back obtained in step (5) are combined The binary image Binary is dot-multiplied, the background information is deleted, and the back grayscale image Gray3 is obtained.

Further, in step (7), the back grayscale image Gray3 that step (6) obtains subtracts image grayscale base value and specifically comprises the following steps:

(71) according to formula Find the mean value of the gray value of each row of the image

Average(i,1), set the gray base value of each row to 0.5Average(i,1);

(72) according to formula Gray4 (i, j)=Gray3 (i, j)-0.5Average (i, 1) the Gray3 image that step (6) obtains every line pixel point gray value subtracts step (71) to obtain The grayscale base value is 0.5Average(i,1), and a new grayscale image Gray4 is obtained.

Further, step (8) obtains the gray-scale median pixel feature point row by row to the new gray-scale image that step (7) obtains specifically comprises the following steps:

(81) creating a new resolution size is a blank binary image Binary2 of n * m, Binary2 (i, j)=0;

(82) For the new grayscale image Gray4 obtained in step (7) according to the formula Perform row summation to get Sum(i,1);

(83) Assign a value to the binary image Binary2, if the new grayscale image Gray4 obtained in step (7) is according to the formula The sum of the pixel points added from left to right by row is within 0.5Average(i,1) of the half of the sum of each row obtained in step (82) 0.5Sum(i,1), then Binary2(i,k) =1, finally get the feature point image of the back contour of the spine.

Further, in step (9), fitting the spine back contour feature points obtained in step (8) into a spine back contour characteristic curve specifically includes the following steps:

(91) Convert the feature points obtained in step (8) from the image coordinate system to the actual common coordinate system, point by point according to the formula Point(k).x=Point(k).i, Point(k).y=Point( k).j convert the feature points from the image coordinate system to the actual common coordinate system;

(92) Fitting the characteristic points of the actual common coordinate system obtained in step (91) into the characteristic curve of the spine contour of the human body, and the fitting method is polynomial fitting by the least squares method.

Beneficial effects: Compared with the prior art, the present invention adopts white light image acquisition, which is non-destructive and radiation-free, fast and convenient, real-time detection, and algorithm is easy to implement, and can effectively extract the back contour curve of the spine to complete the task of scoliosis inspection.

Description of drawings

Fig. 1 is a schematic flow sheet of the method of the present invention;

Figure 2 is an image of the back of a human body obtained by a digital camera;

Figure 3 is a grayscale image of the back of a human body;

Figure 4 is a grayscale image of the red channel;

Fig. 5 is a binary contour image of the back of a human body;

Fig. 6 is the back gray scale image that removes background interference;

Fig. 7 is the grayscale image of subtracting image grayscale base value;

Fig. 8 is the feature point image of the spine back contour;

Fig. 9 is an image of the characteristic curve of the back contour of the spine.

Detailed ways

The technical solution of the present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

In order to reduce the waste of labor and money, improve the detection efficiency of scoliosis, and avoid the false detection caused by the subjective factors of physical measurement, the scoliosis back contour curve extraction method based on the gray value median can indirectly outline the spine contour curve from the body surface. It is a non-destructive, radiation-free, simple and effective scoliosis inspection method to judge the degree of scoliosis.

As shown in Figure 1, a method for extracting scoliosis back contour curve based on gray-scale median value comprises the following steps:

(1) Arrange blue background cloth and photographic lighting to ensure that the light is evenly irradiated on the back of the human body;

(2) The back image of the human body is captured by a digital camera, and the color image information of the back of the upright human body is collected, including RGB three-channel brightness information and image resolution size n×m; the color image of the back as shown in Figure 2 is obtained. When collecting color images of the back of the human body, the person to be collected needs to stand upright, and the image includes the uppermost end of the cervical spine to the lowermost end of the lumbar spine.

(3) grayscale processing is carried out to the color image of the back of the human body acquired in step (2), to obtain a grayscale image;

According to the following formula to the color image that step (2) obtains, carry out the RGB three-component weighted mean processing and obtain the grayscale image Gray1 shown in Figure 3;

Gray1(i,j)=0.30R(i,j)+0.59G(i,j)+0.11B(i,j);

Among them, R(i,j), G(i,j), and B(i,j) are the brightness values of red, green, and blue three-channel component pixels, respectively, and Gray1(i,j) is the obtained grayscale image pixel gray degree value;

(4) Carry out channel separation to the color image of the back of the human body acquired in step (2), and obtain the grayscale image information of the red channel;

Separate the RGB three-component channel on the color image of the back of the human body obtained in step (2), and extract the red channel image information separately according to the formula Gray2(i,j)=R(i,j), and obtain the red channel as shown in Figure 4 In the grayscale image Gray2, R(i,j) is the brightness value of the pixel of the red channel component, and Gray2(i,j) is the grayscale value of the pixel of the grayscale image of the red channel.

(5) Convert the red channel grayscale image Gray2 in step (4) into a binary image to obtain a binary image of the back of the human body; specifically include the following sub-steps:

(51) Set the gray scale threshold T;

(52) Binarize the red channel grayscale image Gray2 obtained in step (4) according to the grayscale threshold T set in (51), if Gray2(i,j)>T, then Binary(i,j) =1; if Gray2(i,j)≦T, then Binary1(i,j)=0, to obtain the binary image Binary1 of the back contour of the human body as shown in FIG. 5 .

(6) dot product the human body back grayscale image in step (3) and the human body back binary image in step (5), obtain the back grayscale image that removes background interference;

According to the formula Gray3(i,j)=Gray1(i,j)*Binary1(i,j), the grayscale image Gray1 obtained in step (3) and the binary image Binary1 of the human back contour obtained in step (5) are dot-multiplied , delete the background information, and obtain the grayscale image of the back that removes the background interference as shown in Figure 6;

(7) Subtract the image grayscale base value from the back grayscale image Gray3 in step (6) to obtain a new grayscale image Gray4, which specifically includes the following steps:

(71) According to formula Calculate the average gray value of each row of the image, Average(i,1), and set the gray value of each row as 0.5Average(i,1);

(72) according to formula Gray4 (i, j)=Gray3 (i, j)-0.5Average (i, 1) the Gray3 image that step (6) obtains every line pixel point gray value subtracts step (71) to obtain The grayscale base value is 0.5Average(i,1), and a new grayscale image Gray4 as shown in FIG. 7 is obtained.

(8) Obtain gray-scale median pixel feature points line by line to the new gray-scale image Gray4 that step (7) obtains, specifically comprise the following steps:

(81) creating a new resolution size is a blank binary image Binary2 of n * m, Binary2 (i, j)=0;

(82) For the new grayscale image Gray4 obtained in step (7) according to the formula Perform row summation to get Sum(i,1);

(83) Assign a value to the binary image Binary2, if the new grayscale image Gray4 obtained in step (7) is according to the formula The difference between the sum of pixels added from left to right by row and the half of the sum of each row obtained in step (82) is within 0.5Average(i, 1), then Binary2(i, k)=1, and finally get as shown in Figure 8 shows the feature point image of the back contour of the spine.

(9) the spine back contour characteristic point that step (8) obtains is fitted into the spine back contour characteristic curve, specifically comprising the following steps:

(91) Convert the feature points obtained in step (83) from the image coordinate system to the actual common coordinate system, point by point according to the formula Point(k).x=Point(k).i, Point(k).y=Point( k).j convert the feature points from the image coordinate system to the actual common coordinate system;

(92) Fitting the feature points of the actual common coordinate system obtained in step (91) into the characteristic curve of the spine contour of the human back, as shown in Figure 9, the fitting method is polynomial fitting by least squares method.

A method for extracting the scoliosis back contour curve based on the gray value median of the present invention, first arranges the blue background cloth and photographic lighting, adopts the ordinary digital camera to take the image of the back of the human body upright, and then extracts the red channel information of the RGB image, Binarize the back contour of the human body, then grayscale the back image and multiply it with the binarized human back contour map to obtain a grayscale image that removes background interference, and then subtract the average grayscale value of the pixels from the grayscale image A new grayscale image is obtained, and then the grayscale median pixel feature points are calculated line by line, and then the feature points are polynomially fitted by the least square method to obtain the spine contour fitting curve. The method of the invention is simple and practical, and the algorithm is easy to realize, and can effectively complete the purpose of extracting the contour curve of the back spine of the human body.

Claims (9)

1. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value, which comprises the following steps:

(1) arrange that blue background cloth and photography light, headlamp need uniform irradiation in human body back；

(2) human body back image is obtained using digital camera shooting, acquires upright human body back color image information, including RGB Triple channel luminance information and image resolution ratio size n × m；

(3) gray processing processing is carried out to the human body back color image obtained in step (2), obtains gray level image；

(4) channel separation is carried out to the human body back color image obtained in step (2), obtains red channel gray level image letter Breath；

(5) the red channel gray level image in step (4) is subjected to binary conversion treatment, obtains human body back bianry image；

(6) the human body back gray level image in step (3) and the human body back bianry image in step (5) are subjected to dot product, obtained To the back gray level image for removing background interference；

(7) by the back gray level image subtracted image gray scale base value in step (6), new gray level image is obtained；

(8) the new gray level image obtained to step (7) seeks gray scale median pixel characteristic point line by line；

(9) characteristic point that step (8) obtains is fitted to backbone back contour feature curve.

2. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, when being acquired in step (2) to human body back color image, is collected people and needs to stand just, image includes from cervical vertebra Topmost arrive lumbar vertebrae bottom.

3. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, the human body back color image obtained in step (3) to step (2) is according to formula Gray1 (i, j)=0.30R (i, j) + 0.59G (i, j)+0.11B (i, j) carries out RGB three-component weighted average and handles to obtain gray level image Gray1, R (i, j), G (i, J), B (i, j) is RGB triple channel component pixel brightness value, the gray level image pixel that Gray1 (i, j) is respectively Gray value.

4. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, RGB three-component channel separation is carried out to the human body back color image that step (2) obtains in step (4), according to formula The independent extraction red channel image information of Gray2 (i, j)=R (i, j), which obtains red channel gray level image Gray2, R (i, j), is Red channel component pixel point brightness value, Gray2 (i, j) are red channel gray level image pixel gray values.

5. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, in step (5) by red channel gray level image Gray2 be converted into bianry image specifically include it is following step by step:

(51) gray threshold T is set；

(52) binaryzation is carried out according to the gray threshold T that (51) are set to the red channel gray level image Gray2 that step (4) obtains Processing, if Gray2 (i, j) > T, Binary (i, j)=1；If Gray2 (i, j)≤T, Binary (i, j)=0, obtain people Body back profile bianry image Binary1.

6. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, the grayscale image for obtaining step (3) according to formula Gray3 (i, j)=Gray1 (i, j) * Binary (i, j) in step (6) As the human body back bianry image Binary progress dot product that Gray1 and step (5) obtain, background information is deleted, back ash is obtained Spend image Gray3.

7. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, specifically includes the back gray level image Gray3 subtracted image gray scale base value that step (6) obtains in step (7) following Step:

(71) according to formulaThe every row gray value mean value Average of image (i, 1) is sought, Every row gray scale base value is set as 0.5Average (i, 1)；

(72) the Gray3 image for obtaining step (6) according to formula Gray4 (i, j)=Gray3 (i, j) -0.5Average (i, 1) Every row pixel gray value subtracts the gray scale base value 0.5Average (i, 1) that step (71) obtains, and obtains new gray level image Gray4。

8. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, the new gray level image that step (8) obtains step (7) seek line by line gray scale median pixel characteristic point specifically include with Lower step:

(81) blank bianry image Binary2, Binary2 (i, j)=0 that the new resolution sizes of one width of creation are n × m；

(82) the new gray level image Gray4 obtained to step (7) is according to formulaRow is carried out to ask With obtain Sum (i, 1)；

(83) assignment is carried out to bianry image Binary2, if the new gray level image Gray4 that step (7) obtains is according to formulaPixel sum and step in rows from left to right (82) half of the sum of the every row obtained 0.5Sum (i, 1) differs within 0.5Average (i, 1), then Binary2 (i, k)= 1, finally obtain backbone back contour feature point image.

9. a kind of scoliosis back contour curve extracting method based on gray scale intermediate value according to claim 1, special Sign is, it is specific that the backbone back contour feature point that step (8) obtains is fitted to backbone back contour feature curve by step (9) The following steps are included:

(91) characteristic point that step (8) obtains is transformed into practical Common Coordinate from image coordinate system, point by point according to formula Point (k) characteristic point is transformed into practical common seat from image coordinate system by .x=Point (k) .i, Point (k) .y=Point (k) .j Mark system；

(92) the practical Common Coordinate characteristic point for obtaining step (91) is fitted to human body back backbone contour feature curve, intends Conjunction mode is least square method fitting of a polynomial.