CN108648229B - Human back feature point extraction method based on Kinect camera - Google Patents

Abstract

The invention discloses a human back feature point extraction method based on a Kinect camera, which comprises the steps of obtaining a human back image, graying, contrast stretching, binaryzation, connected domain marking, traversing and searching for human back armpit, shoulder, waist and crotch feature points, and marking in the human back image. According to the invention, by extracting the outline image of the human back image and traversing the outline image, the characteristic points of the human back image can be efficiently and accurately extracted, and accurate data is provided for the subsequent calculation of the scoliosis degree of the human body.

Description

Human back feature point extraction method based on Kinect camera

technical field

The invention belongs to the field of digital image processing, in particular to a method for extracting feature points of the back of a human body based on a Kinect camera.

Background technique

Scoliosis can also be called scoliosis, which has multiple characteristics. In recent years, the incidence of scoliosis in adolescents is increasing year by year, which has a certain degree of impact on adolescents' lives and future work. Therefore, scoliosis detection has quite a wide range of applications. The commonly used methods for scoliosis are the scoliosis ruler and the Adams forward bending test, but these detection methods are not accurate enough, and relying on manual detection is not only laborious, but also the results obtained are not accurate enough, which may easily lead to false detection and missed detection. The automatic feature extraction based on digital image processing can ensure the accuracy and save a lot of manpower and material resources.

SUMMARY OF THE INVENTION

The purpose of the present invention is to: in order to solve the above problems existing in the prior art, the present invention proposes a method for extracting feature points of the back of a human body based on a Kinect camera.

The technical scheme of the present invention is: a method for extracting feature points on the back of a human body based on a Kinect camera, comprising the following steps:

A. Obtain the image of the back of the human body, and perform grayscale processing on the image of the back of the human body;

B. Contrast stretching is performed on the grayscale image obtained in step A;

C, adopt the Otsu threshold binarization method to process the image processed in step B to obtain a binary image;

D, using the connected domain method to process the binary image obtained in step C to obtain a binary image of the back of the human body;

E, adopt the canny operator method to process the binary image of the back of the human body obtained in step D, and obtain the outline of the back of the human body;

F. Traverse the outer contour map of the back of the human body obtained in step E, and obtain the left armpit coordinate point, right armpit coordinate point, left shoulder coordinate point, right shoulder coordinate point, waist coordinate point and hip bone of the back of the human body respectively. Coordinate points;

G. Mark the multiple coordinate points obtained in step F in the image of the back of the human body in step A.

Further, the step D uses the connected domain method to process the binary image obtained in step C to obtain a binary image of the back of the human body. Specifically, the connected domain is marked on the binary image obtained in step C, and the statistics of the connected regions are calculated. Then calculate the area of each connected area for the obtained connected area, set the connected area smaller than the maximum connected area area to zero, and obtain a binary image of the back of the human body.

Further, the step D further includes performing hole filling processing on the obtained binary image of the back of the human body.

Further, the step F includes calculating the center coordinates and the back contour coordinates (i ₁ , k ₁ ) and (i ₂ , k ₂ ) on the left and right sides of the i-th row of the human body back outline obtained in the step E.

Further, the step F traverses the outline of the back of the human body, and obtains the coordinates of the left armpit of the back of the human body. Specifically: traverse the outline of the back of the human body from the jth column of the i _1-1 row to find the human body. The column coordinate k′ ₁ of the left contour of the back is used to determine whether the difference k ₁ between the column coordinate k′ ₁ and the column coordinate is greater than the set threshold _; The lower coordinate point (m ₁ , n ₁ ); if not, continue to traverse the i ₁ -2 row.

Further, the step F traverses the outline of the back of the human body, and obtains the right armpit coordinate point of the back of the human body. Specifically, the outline of the back of the human body is traversed from the jth column of the _i2-1 row to find the human body. The column coordinate k' ₂ of the right contour of the back is used to determine whether the difference k ₂ between the column coordinate k' ₂ and the column coordinate is greater than the set threshold _; The lower coordinate point (m ₂ , n ₂ ); if not, continue to traverse the i ₂ -2th row.

Further, the step F traverses the outline of the back of the human body, and obtains the left shoulder coordinate point of the back of the human body. Specifically: according to the outline of the back of the human body, traverse n ₁ columns starting from the m ₁ -1 row, and find a The zero point is used as the coordinate point of the left shoulder of the back of the human body.

Further, the step F traverses the outline of the back of the human body, and obtains the right shoulder coordinate point of the back of the human body. Specifically: according to the outline of the back of the human body, traverse n ₂ columns starting from the m ₂ -1 row, and find the The zero point is taken as the coordinate point of the right shoulder of the back of the human body.

Further, the step F traverses the outline of the back of the human body, and obtains the coordinates of the waist and the crotch of the back of the human body. Specifically, the corner detection algorithm based on the curvature is used to detect the outline of the back of the human body, including the following parts: step:

S1. Starting from the center row coordinates of the outer contour map of the back of the human body, obtain the contour coordinates of the left and right sides of the back of the human body respectively;

S2. Starting from the Nth contour point of the contour coordinates on the left and right sides of the back of the human body, select this point as the current contour point p _i in turn, and the contour point separated from the current contour point by N-1 points is the front contour point p _{i-( N-1)} and back contour point p _i+(N-1) ;

S3. Calculate the curvature of the current contour point p _i according to the curvature formula;

S4. Find the contour points corresponding to the maximum and minimum curvatures from the curvatures of the contour points on the left and right sides of the back of the human body, which are the coordinate points of the crotch and the waist of the back of the human body.

Further, the curvature formula in the step S3 is specifically:

Among them, k(i) is the curvature of the ith contour point, |pi p _ik | is the distance between the current contour point and the previous contour point at k points, |pi p _{i +k |} is the current contour point and its distance The distance between the back contour points separated by k points, |p _ik p _i+k | is the distance between the front contour point p _ik and the back contour point p _i+k .

The beneficial effects of the present invention are as follows: the present invention can extract the feature points of the back image of the human body efficiently and accurately by extracting the outer contour of the image of the back of the human body and traversing the outer contour, thereby providing accurate and accurate results for the subsequent calculation of the degree of scoliosis of the human body The data.

Description of drawings

1 is a schematic flowchart of a method for extracting feature points on the back of a human body based on a Kinect camera of the present invention;

2 is a grayscale image after grayscale processing in an embodiment of the present invention;

Fig. 3 is the binary image after adopting Otsu threshold binarization in the embodiment of the present invention;

4 is a schematic diagram of the outer contour of the back of the human body in the embodiment of the present invention;

FIG. 5 is a schematic diagram of an image labeling result of the back of a human body in an embodiment of the present invention.

Detailed ways

In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention, but not to limit the present invention.

As shown in FIG. 1 , it is a schematic flowchart of the method for extracting feature points of human back based on the Kinect camera of the present invention. A method for extracting human back feature points based on a Kinect camera, comprising the following steps:

A. Obtain the image of the back of the human body, and perform grayscale processing on the image of the back of the human body;

B. Contrast stretching is performed on the grayscale image obtained in step A;

C, adopt the Otsu threshold binarization method to process the image processed in step B to obtain a binary image;

D, using the connected domain method to process the binary image obtained in step C to obtain a binary image of the back of the human body;

E, adopt the canny operator method to process the binary image of the back of the human body obtained in step D, and obtain the outline of the back of the human body;

F. Traverse the outer contour map of the back of the human body obtained in step E, and obtain the left armpit coordinate point, right armpit coordinate point, left shoulder coordinate point, right shoulder coordinate point, waist coordinate point and hip bone of the back of the human body respectively. Coordinate points;

G. Mark the multiple coordinate points obtained in step F in the image of the back of the human body in step A.

In an optional embodiment of the present invention, in the above step A, a Kinect camera is used to obtain an image of the back of the human body, and the size of the image is 1920×1080; and grayscale processing is performed on the obtained image of the back of the human body. As shown in FIG. 2, it is a grayscale image after grayscale processing in the embodiment of the present invention; as shown in FIG. 3, it is a binary image after binarization using Otsu threshold in the embodiment of the present invention; as shown in FIG. 4 It is a schematic diagram of the outer contour of the back of the human body in the embodiment of the present invention.

In an optional embodiment of the present invention, the above-mentioned step D adopts the connected domain method to process the binary image obtained in step C to obtain a binary image of the back of the human body. Specifically, the connected domain is performed on the binary image obtained in step C. Mark, and count the number of connected regions; then calculate the area of each connected region for the obtained connected regions, filter, and set the connected regions smaller than the largest connected region to zero to obtain a binary image of the back of the human body.

Since the color of the back image of the human body may be different, some holes will be generated after binarization, so the hole filling process is performed on the obtained binary image of the back of the human body.

In an optional embodiment of the present invention, the above-mentioned step F first calculates the center coordinates (i,j) and the back contour coordinates (i ₁ , k ₁ ) and (i ₂ , k ₂ ).

Traverse the outline of the back of the human body, and obtain the coordinates of the left armpit of the back of the human body. Specifically, the outline of the back of the human body is traversed from the i ₁ -1 row upwards, and the jth column of the i ₁ -1 row is traversed. Start to traverse to the 0th column, find the column coordinate k' ₁ of the left contour of the back of the human body, and judge whether the difference between the column coordinate k' ₁ and the column coordinate k ₁ is greater than the set threshold, and the threshold here is set to 10; if so, then Take the coordinates of the left contour of row i ₁ as the coordinate point (m ₁ , n ₁ ) of the left armpit of the back of the human body; if not, continue to traverse rows i ₁ -2 to find the new column of the left contour of the back of the human body coordinate, and determine whether the difference between the column coordinate and the column coordinate k′ ₁ is greater than the set threshold; if so, take the left contour coordinate of the i ₁ -1 row as the coordinate point of the left armpit of the back of the human body; iterate according to this method Execute until the coordinate point of the armpit on the left side of the back of the human body is found.

Traverse the outer contour map of the back of the human body to obtain the coordinate points of the right armpit of the back of the human body. Specifically: traverse the outer contour map of the back of the human body from the jth column of row i ₂ -1 to find the column coordinates of the outline of the right side of the human back. k' ₂ , judge whether the difference between the column coordinate k' ₂ and the column coordinate _{k 2} is greater than the set threshold _; n ₂ ); if not, then continue to traverse the i ₂ -2 row, find the column coordinates of the left contour of the new back of the human body, and judge whether the difference between the column coordinates and the column coordinates k′ ₂ is greater than the set threshold; if so, Then, the coordinates of the right contour of the i ₂ -1 line are taken as the coordinate point of the right armpit on the back of the human body; the method is iteratively executed until the coordinate point of the right armpit on the back of the human body is found.

Traverse the outer contour map of the back of the human body, and obtain the coordinate point of the left shoulder of the back of the human body. Specifically: according to the outer contour map of the human back, traverse the n ₁ columns of the outer contour map of the human back from the m ₁ -1 row upwards, and find no The zero point is used as the coordinate point of the left shoulder of the back of the human body.

Traverse the contour map of the back of the human body, and obtain the coordinate point of the shoulder on the right side of the back of the human body. Specifically: according to the contour map of the back of the human body, traverse _n2 columns of the contour map of the back of the human body from the m ₂ -1 row upwards, and find no The zero point is used as the shoulder coordinate point on the right side of the back of the human body.

Traverse the outer contour map of the back of the human body to obtain the waist coordinate points and the coordinate points of the hip bone on the back of the human body. Specifically, the corner detection algorithm based on the curvature of the human body is used to detect the outer contour of the back of the human body, including the following sub-steps:

S1. Starting from the center row coordinates of the outer contour map of the back of the human body, obtain the contour coordinates of the left and right sides of the back of the human body respectively;

S2. Starting from the Nth contour point of the contour coordinates on the left and right sides of the back of the human body, select this point as the current contour point p _i in turn, and the contour point separated from the current contour point by N-1 points is the front contour point p _{i-( N-1)} and back contour point p _i+(N-1) ;

The present invention starts from the 5th contour point of the contour coordinates on the left and right sides of the back of the human body, and sequentially selects this point as the current contour point p _i , and the contour points separated from the current contour point by 4 points are the front contour points p _i-4 and Back contour point p _i+4 .

S3. Calculate the curvature of the current contour point p _i according to the curvature formula, and the curvature formula is specifically:

Among them, k(i) is the curvature of the ith contour point, |pi p _ik | is the distance between the current contour point and the previous contour point at k points, |pi p _{i +k |} is the current contour point and its distance The distance between the back contour points separated by k points, |p _ik p _i+k | is the distance between the front contour point p _ik and the back contour point p _i+k .

S4. Find the contour points corresponding to the maximum and minimum curvatures from the curvatures of the contour points on the left and right sides of the back of the human body, which are the coordinate points of the crotch and the waist of the back of the human body. As shown in FIG. 5 , it is a schematic diagram of an image labeling result of the back of the human body in the embodiment of the present invention.

Those of ordinary skill in the art will appreciate that the embodiments described herein are intended to assist readers in understanding the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations without departing from the essence of the present invention according to the technical teaching disclosed in the present invention, and these modifications and combinations still fall within the protection scope of the present invention.

Claims (4)

1. A human back feature point extraction method based on a Kinect camera is characterized by comprising the following steps:

A. acquiring a back image of a human body, and carrying out gray processing on the back image of the human body;

B. b, performing contrast stretching treatment on the gray level image obtained in the step A;

C. b, processing the image processed in the step B by adopting an Otsu threshold binarization method to obtain a binary image;

D. c, processing the binary image obtained in the step C by adopting a connected domain method to obtain a human back binary image;

E. d, processing the human body back binary image obtained in the step D by adopting a canny operator method to obtain a human body back outline image;

F. e, traversing the human back outline diagram obtained in the step E, and respectively obtaining a left-side axillary coordinate point, a right-side axillary coordinate point, a left-side shoulder coordinate point, a right-side shoulder coordinate point, a waist coordinate point and a crotch coordinate point of the human back, specifically:

calculating the center coordinates of the human body back outline drawing obtained in the step E and the back outline coordinates (i) at the left side and the right side of the ith row of the outline drawing₁,k₁) And (i)₂,k₂)；

From the ith to the human back outline₁Starting to traverse the jth column of the-1 row, and finding the column coordinate k of the left outline of the back of the human body₁', judging the column coordinate k₁' and column coordinate k₁Whether the difference is greater than a set threshold value; if yes, the ith₁The left contour coordinate of the row is taken as the left axillary coordinate point (m) of the back of the human body₁,n₁) (ii) a If not, continue traversing ith₁-2 rows;

from the ith to the human back outline₂Starting to traverse the jth column of the 1 row, and finding the column coordinate k of the contour of the right side of the back of the human body₂', judging the column coordinate k₂' and column coordinate k₂Whether the difference is greater than a set threshold value; if yes, the ith₂The right contour coordinate of the line is taken as the coordinate point (m) of the right armpit of the back of the human body₂,n₂) (ii) a If not, continue traversing ith₂-2 rows;

according to the outline diagram of the back of the human bodyM th₁1 line starts traversing n₁Column, finding out a point which is not zero as a coordinate point of the left shoulder of the back of the human body;

according to the outer contour diagram of the back of the human body from the m-th₂1 line starts traversing n₂Column, finding out a point which is not zero as a coordinate point of the right shoulder of the back of the human body;

the method for detecting the corners of the human body back outline by adopting the corner detection algorithm based on curvature comprises the following steps:

s1, respectively acquiring contour coordinates of the left side and the right side of the back of the human body from the center line coordinate of the outline drawing of the back of the human body;

s2, sequentially selecting the Nth contour point of the contour coordinates of the left side and the right side of the back of the human body as the current contour point p_iThe contour point spaced from the current contour point by N-1 points is a front contour point p_i-(N-1)And back contour point p_i+(N-1)；

S3, calculating the current contour point p according to the curvature formula_iThe curvature of (a);

s4, finding out contour points corresponding to the maximum value and the minimum value of curvature from the curvatures of the contour points on the left side and the right side of the back of the human body, namely the hip coordinate points and the waist coordinate points of the back of the human body;

G. and D, marking the human back image in the step A according to the plurality of coordinate points obtained in the step F.

2. The method for extracting human back feature points based on the Kinect camera as claimed in claim 1, wherein the step D adopts a connected domain method to process the binary image obtained in the step C to obtain a human back binary image, specifically: c, carrying out connected domain marking on the binary image obtained in the step C, and counting the number of connected domains; and calculating the area of each connected region for the obtained connected regions, and setting the connected region smaller than the area of the maximum connected region to zero to obtain a binary image of the back of the human body.

3. The method as claimed in claim 2, wherein the step D further comprises performing hole filling processing on the obtained binary image of the back of the human body.

4. The method for extracting human back feature points based on a Kinect camera as claimed in claim 1, wherein the curvature formula in the step S3 is specifically:

where k (i) is the curvature of the ith contour point, | p_ip_i-kI is the distance between the current contour point and the front contour point spaced by k points, | p_ip_i+kI is the distance between the current contour point and the back contour point separated by k points, | p_i-kp_i+kI is the front contour point p_i-kAnd the back contour point p_i+kThe distance of (c).

Images