A method for detecting the precise contour of the foot in standing long jump

Abstract

The present invention proposes a method for detecting the precise contour of a foot in a standing long jump. The present invention belongs to the field of precise edge detection, and in particular, relates to a method for detecting the precise contour of a foot in a standing long jump using contour detection technology and shadow elimination technology. The method comprises the following steps: 1. Acquire a take-off image and a landing image; 2. Acquire a circumscribed rectangle of the landing foot; 3. Add a landing foot contour buffer; 4. Detect the direction information of the landing foot contour; 5. Detect the precise edge contour of the foot. The present invention uses a contour detection method of two frames of images, firstly detecting the foot contour in the image of the standing long jump take-off moment, and then detecting the precise contour of the foot by performing a differential with the foot grayscale image of the standing long jump landing moment image. The present invention particularly greatly improves the contour detection error caused by shadows.

Description

Technical Field

[0001] The invention belongs to the field of precise edge detection, and in particular relates to a method for detecting the precise contour of a standing long jump foot by using contour detection technology and shadow elimination technology. Background Art

[0002] In the standing long jump test using machine vision technology, it is necessary to obtain accurate foot contours as a basis for performance measurement and judgment of action compliance. For example, in the standing long jump test, it is necessary to obtain the accurate foot contour position, and then obtain the accurate heel position as the landing point for measuring the performance, and then calculate the vertical distance between the landing point and the take-off line as the standing long jump score.

[0003] The current common contour detection method can only detect the approximate foot contour. At the same time, the lighting has a great impact on contour detection. Usually, the detection method will regard the heel shadow as part of the foot. Therefore, the common contour detection method cannot be used as the contour basis for measuring the performance in standing long jump. Summary of the invention

[0004] The present invention uses a two-frame image contour detection method, firstly detecting the foot contour in the image of the moment of standing long jump landing, and then detecting the accurate foot contour by performing a differential with the foot grayscale image of the moment of standing long jump take-off. The present invention can greatly improve the contour detection error caused by shadows.

[0005] The present invention is implemented by adopting the following technical solutions.

[0006] A method for detecting the precise contour of a foot in a standing long jump, comprising:

[0007] Step 1: Get the take-off image and landing image

[0008] Get two frames of standing long jump images, one frame is the image before take-off, and the other frame is the image at the moment of landing. These two frames are taken by the same camera, so the image size and background are the same.

[0009] Step 2, obtain the circumscribed rectangle of the landing foot.

[0010] The landing image is processed and analyzed, and the minimum circumscribed rectangle of the left foot is obtained as the left foot rectangle, and the minimum circumscribed rectangle of the right foot is obtained as the right foot rectangle.

[0011] Step 3: Add a buffer zone for the foot contour when landing.

[0012] Take the left foot rectangle, increase the width of the left foot rectangle by a certain amount on the left and right, increase the height of the rectangle by a certain amount downward, and define the rectangle with the added buffer as the left foot buffer outline. Similarly, obtain the right foot buffer outline.

[0013] Step 4: Detect the contour direction information of the landing foot

[0014] The grayscale approximation of the brightness of the foot buffer contour image is calculated to obtain the edge contour direction information of the foot. The left foot contour image with edge direction information is obtained as GL1, and the right foot contour image is obtained as GR1.

[0015] Step 5: Detect the precise edge contour of the foot

[0016] Take the position (x1, x2, y1, y2) of the image GL1 in the original image (i.e. the landing image), cut out the picture at the same position in the take-off image, and take the grayscale image, which is defined as the image GL2.

[0017] Difference is performed between GL1 and GL2, and the contour obtained after the difference is the precise edge contour of the left foot.

[0018] Repeat the same process for GR1 to obtain the exact edge contour of the right foot. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A schematic flow chart of a method for detecting the precise outline of a foot in standing long jump provided by an embodiment of the present invention is shown.

[0020] Figure 2 A schematic diagram of generating an outline rectangle of a method for detecting an accurate outline of a foot in standing long jump provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0021] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0022] This embodiment discloses a method for detecting the precise contour of the foot in standing long jump, please refer to Figure 1 ,include:

[0023] Step S101, obtaining take-off image and landing image

[0024] Get two frames of standing long jump images, one frame is the image before take-off, and the other frame is the image at the moment of landing. These two frames are taken by the same camera, so the image size and background are the same.

[0025] Step S102: Obtain the circumscribed rectangle of the landing foot

[0026] The landing image is processed and analyzed, and the key nodes of the foot are obtained using a convolutional neural network: left ankle A, right ankle B, left heel C, right heel D, left little toe E, right little toe F, left big toe G, and right big toe H. Take the left foot as an example, refer to the attached Figure 2 , A represents the left ankle, C represents the left heel, E represents the left little toe, and G represents the left big toe.

[0027] Take A, C, E, G as the left foot group, B, D, F, H as the right foot group, take the smallest circumscribed rectangle of the left foot group as the left foot rectangle, and take the smallest circumscribed rectangle of the right foot group as the right foot rectangle. Figure 2 .

[0028] Step S103: Adding a foot landing contour buffer

[0029] Take the left foot rectangle and get the four vertices of the rectangle. Increase the width of the left foot rectangle by CE / 2 (half the distance from the heel to the big toe) on both sides, increase the height of the rectangle by EG (the distance from the big toe to the little toe), and define the rectangle with the added buffer as the left foot buffer outline. See the attached Figure 2 . The same goes for the right foot cushioning profile.

[0030] Step S104: Detect the contour direction information of the landing foot

[0031] The Sobel discrete convolution factor is used to calculate the grayscale approximation of the brightness of the foot buffer contour image to obtain the edge contour direction information of the foot.

[0032] The Sobel convolution factor contains two sets of 3x3 matrices, namely the horizontal convolution factor and the vertical convolution factor. The horizontal and vertical brightness difference approximations are obtained by plane convolution with the image. The horizontal and vertical convolution factors are:

[0033]

[0034] If Raw represents the original image, Gx and Gy represent the grayscale values ​​of the image after horizontal and vertical edge detection respectively, the formula is as follows:

[0035]

[0036] The specific calculation is as follows:

[0037] Gx = (-1)*f(x-1, y-1) + 0*f(x,y-1) + 1*f(x+1,y-1)+(-2)*f(x-1,y) + 0*f(x,y)+2*f(x+1,y)+(-1)*f(x-1,y+1) +0*f(x,y+1) + 1*f(x+1,y+1) = [f(x+1,y-1)+2*f(x+1,y)+f(x+1,y+1)]-[f(x-1,y-1)+2*f(x-1,y)+f(x-1,y+1)];

[0038] Gy =1* f(x-1, y-1) + 2*f(x,y-1)+ 1*f(x+1,y-1)+0*f(x-1,y) 0*f(x,y) +0*f(x+1,y)+(-1)*f(x-1,y+1) +(-2)*f(x,y+1) + (-1)*f(x+1, y+1) = [f(x-1,y-1) +2f(x,y-1) + f(x+1,y-1)]-[f(x-1, y+1) + 2*f(x,y+1)+f(x+1,y+1)];

[0039] Where f(a,b) represents the grayscale value of the image point (a,b).

[0040] The grayscale value of each pixel in the image is calculated using the following formula, taking the horizontal and vertical grayscale values ​​as parameters:

[0041] .

[0042] If higher computational efficiency is required, use the approximate calculation formula without square root extraction:

[0043] .

[0044] If G is greater than a certain threshold, the point (x, y) is considered to be an edge point.

[0045] By performing the above Sobel operator convolution grayscale processing on the image, the edge direction information of the image is obtained. The above calculation is performed on the left foot buffer contour image and the right foot buffer contour image respectively, and the left foot contour image GL1 and the right foot contour image GR1 with edge direction information are obtained.

[0046] Step S105: Detect the precise edge contour of the foot

[0047] Take the position (x1, x2, y1, y2) of the image GL1 in the original image (i.e. the landing image), cut out the picture at the same position in the take-off image, and take the grayscale image, which is defined as the image GL2.

[0048] Difference is performed between GL1 and GL2, and the contour obtained after the difference is the precise edge contour of the left foot.

[0049] Repeat the same process for GR1 to obtain the exact edge contour of the right foot.

[0050] Obviously, the above embodiments are merely examples for clearly explaining the present invention, and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived therefrom are still within the protection scope of the present invention.

Claims

1. A method for detecting the precise contour of a foot in a standing long jump, characterized in that:

1. Get the take-off image and the landing image; 2. Get the circumscribed rectangle of the landing foot; 3. Add the contour buffer of the landing foot; 4. Detect the contour direction information of the landing foot; 5. Detect the precise edge contour of the foot; The detecting of the contour direction information of the landing foot comprises: calculating the grayscale approximate value of the brightness of the foot buffer contour image, and obtaining the foot contour image with edge contour direction information; Among them, the detection of the precise edge contour of the foot includes: taking the position of the foot contour map in the landing image, cutting out the picture of the same position in the take-off image, and generating a grayscale image, and differentiating the grayscale image and the foot contour map with edge contour direction information. The contour in the picture obtained after the difference is the precise edge contour of the foot.

2. A method for detecting the precise outline of a foot in standing long jump according to claim 1, characterized in that: The acquisition of the take-off image and the landing image is to acquire two frames of images during the standing long jump test, which are an image immediately before the take-off and an image immediately after the landing; these two frames of images are taken by the same camera.

3. The method for detecting the precise outline of the foot in standing long jump according to claim 1, characterized in that: The method obtains the circumscribed rectangle of the landing foot, processes and analyzes the landing image, obtains the circumscribed minimum rectangle of the left foot as the left foot rectangle, and obtains the circumscribed minimum rectangle of the right foot as the right foot rectangle.

4. A method for detecting the precise outline of the foot in standing long jump according to claim 1, characterized in that: The described method of increasing the contour buffer of the landing foot is to take the left foot rectangle and obtain the four vertices of the rectangle; increase the width of the left foot rectangle by a certain amount on the left and right, and increase the height of the rectangle by a certain amount downward, and define the rectangle with the increased buffer as the left foot buffer contour; similarly, obtain the right foot buffer contour.

Images