CN110102040B - A method of counting audio skipping rope based on cross-correlation coefficient method - Google Patents

Abstract

The invention discloses an audio skipping rope counting method based on a cross-correlation coefficient method, which belongs to the technical field of intelligent fitness and sports, comprising: 1) acquiring video data of rope skipping actions, and extracting audio data from the video data; Frequency, draw the audio waveform with time as the abscissa; 3) Select the sample sound, use the cross-correlation coefficient method to calculate the number of skipping ropes; 4) Output and display the counting result. High-definition video recording equipment (such as smart phones, etc.) can be used to record the entire rope skipping process, including the rope skipper, from a certain angle and a suitable distance. After a relatively stable video image is obtained, the audio information in it is analyzed and processed. The method judges the number of skipping ropes by recognizing the sound of the rope touching the ground, which is not only accurate, but also the video can be reshaped, which has good application value.

Description

A method of counting audio skipping rope based on cross-correlation coefficient method

technical field

The invention relates to the technical field of intelligent fitness equipment, in particular to an audio skipping rope counting method based on a cross-correlation coefficient method.

Background technique

As an aerobic exercise suitable for all ages, skipping rope can make every joint of the body get a good exercise. At present, although there are skipping ropes that can be automatically counted, the number of times of counting may be more than once in a short period of time, and the inability to correctly judge the direction of the rope, resulting in invalid results in the process of skipping ropes cannot be automatically excluded.

Chinese patent document with publication number CN106512285A discloses a skipping rope and skipping rope counting method, the skipping rope includes a first handle, a second handle and a rope, the first handle includes a hand-held part, a rotating mechanism, a speed sensor, a processor, an infrared emission The hand-held part is provided with a support part for supporting the infrared receiver, and the infrared transmitter emits an infrared beam toward the infrared receiver. Among them, the speed sensor is used to detect the rotational speed of the rotating mechanism, and the processor judges whether the rotating mechanism rotates once according to the rotational speed data of the rotating mechanism. Whether they are consistent, if the directions are the same, the rotational speed data is valid, and if the directions are opposite, the rotational speed data is invalid. Although the present invention can realize the precise counting of the rope skipping process, there are often problems such as insufficient counting accuracy, inability to reverse counting, and long service time.

The Chinese patent document with the publication number of CN109758716A discloses a method for counting skipping ropes based on sound information, which uses the pulse extreme point after the sound filtering generated by the contact between the rope and the ground to calculate the number of skipping ropes. Conditions that cause miscounts.

SUMMARY OF THE INVENTION

The purpose of the present invention is to provide a method for counting skipping ropes based on sound information. The method uses the cross-correlation coefficient method to analyze and judge audio information by analyzing the mechanism of people's use of auditory perception in the process of skipping rope counting. Jump rope for automatic precise counting.

In order to achieve the above object, the audio skipping rope counting method based on the cross-correlation coefficient method provided by the present invention comprises the following steps:

1) obtain the video data of the rope skipping action, and extract audio data from the video data;

2) Extract the sampling frequency of the audio, and draw an audio waveform with time as the abscissa;

3) Select the sample sound, and use the cross-correlation coefficient method to calculate the number of skipping ropes;

4) Output and display the counting result.

In the above technical solution, a high-definition video recording device (such as a smart phone, etc.) can be used to record the entire rope skipping process including the rope skipper from a certain angle and a suitable distance. After a relatively stable video image is obtained, the audio information in it is analyzed and processed. The method determines the number of times of rope skipping by recognizing the sound of the rope touching the ground, which is not only accurate, but also the video can be reshaped, which has good application value.

Preferably, in step 2), the sampling frequency of the audio is extracted, and the step of drawing the audio waveform with time as the abscissa is: extracting the audio information of the rope skipping video as the tested audio to obtain its time-domain waveform.

As preferably, in step 3), selected sample sound, the step that adopts cross-correlation coefficient method to calculate skipping times is:

3-1) Take the audio clip of a single impact between the skipping rope and the ground as the sample audio x, and use the audio sequence in the skipping video as the tested audio y, calculate the cross-correlation coefficient between the sample audio x and the tested audio y, and the cross-correlation The calculation method of the function is explained as follows:

表示序列x保持不动，将序列y左移m个抽样点后，两个序列逐点对应相乘的结果；

表示序列y保持不动，将序列x左移m个抽样点后，两个序列逐点对应相乘的结果。互相关函数是用于表征两个信号x，y在任意两个不同时刻的取值之间的相关程度，它是两个信号x，y在频域内是否相关的一个重要判断标准。互相关系数是对两个信号x，y作互相关运算后的结果进行归一化得到的。互相关系数越大，表明两个信号的相关程度越高。where N is the length of the longer signal sequence in x,y, and the asterisks indicate complex conjugation. m represents the number of sampling points for translation;

Indicates that the sequence x remains unchanged and the sequence y is shifted to the left by m sampling points, and the result of the corresponding multiplication of the two sequences point by point;

Indicates that the sequence y remains unchanged and the sequence x is shifted to the left by m sampling points, and the result of the corresponding multiplication of the two sequences point by point. The cross-correlation function is used to characterize the degree of correlation between the values of two signals x, y at any two different times. It is an important criterion for judging whether two signals x, y are correlated in the frequency domain. The cross-correlation coefficient is obtained by normalizing the result of the cross-correlation operation of the two signals x and y. The larger the cross-correlation coefficient, the higher the correlation between the two signals.

3-2) Selection of sample audio; in the selection of sample audio, the sound clip of the single impact of the skipping rope and the ground is used as the sample audio, and the sound of people landing is not included. The value of the cross-correlation coefficient corresponding to the landing sound is very small, so that the influence of the different sounds produced by different rope skippers on the landing on the final rope skipping count is excluded;

Considering the interference of various random sounds during the rope skipping process, when selecting the sample audio, randomly select n times of skipping rope and the sound clips of a single impact on the ground to perform a weighted average, and n=3 can be taken.

3-3) Draw the "cross-correlation coefficient" graph, set the appropriate cross-correlation coefficient threshold (the threshold can be set to 0.1), and count the number of times the cross-correlation coefficient exceeds the set threshold n, then n is the number of skipping ropes.

Preferably, the minimum data number interval between two skipping rope counts is set. According to experience, there are 300 records of single person skipping rope in 1 minute, that is, each rope skipping period is greater than 0.2s, and the sampling frequency of audio is Fs=44100, that is, the interval between two skipping rope counts should be at least 0.2*44100=8820 pieces of data .

Compared with the prior art, the beneficial effects of the present invention are:

The audio skipping rope counting method based on the cross-correlation coefficient method of the present invention can realize automatic and accurate counting without manual work, and can not only count in real time, but also perform video playback, especially with the widespread adoption of mobile video recording equipment, the method will have more Strong practicality.

Description of drawings

1 is a flowchart of an audio skipping rope counting method according to an embodiment of the present invention;

2 is an audio waveform diagram with time as abscissa in the embodiment of the present invention;

3 is a time-domain waveform diagram of sample audio in an embodiment of the present invention;

FIG. 4 is a cross-correlation coefficient diagram of the sample audio and the tested audio 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 below with reference to the embodiments and the accompanying drawings.

Example

Referring to FIGS. 1 to 4 , the audio skipping rope counting method based on the cross-correlation coefficient method of the present embodiment includes the following steps:

In step S1, the original video data of the rope skipping action is obtained through the video recording device, and the original video data is input to start timing.

Step S2, extracting sound data from the video data, extracting the sampling frequency of the audio, and drawing an audio waveform with time as the abscissa. The time-domain waveform diagram obtained is shown in Figure 2.

In step S3, sample sounds are selected, and the number of skipping ropes is calculated by the cross-correlation coefficient method. First, the audio clips of 3 times of skipping rope and a single impact on the ground are weighted and averaged as the sample audio x, and the time domain waveform is shown in Figure 3. The audio sequence in the skipping video is used as the tested audio y, and the sample audio x is calculated. The cross-correlation coefficient with the measured audio y. Then draw a "cross-correlation coefficient" graph, as shown in Figure 4, set an appropriate cross-correlation coefficient threshold (the threshold is set to 0.1 in this example), and count the number of times the cross-correlation coefficient exceeds the set threshold n, then n is the number of skipping ropes .

Step S4, it is judged whether it is a valid skipping rope. According to experience, there are 300 records of single person skipping rope in 1 minute, that is, each rope skipping period is greater than 0.2s, and the sampling frequency of audio is Fs=44100, that is, the interval between two skipping rope counts should be at least 0.2*44100=8820 pieces of data , only for effective skipping.

Step S5, if it is a valid number of skipping ropes, the count is increased by 1, and if it is not a valid skipping rope, it is not counted.

In step S6, it is judged whether the counting time is exceeded, and if so, the counting is ended; otherwise, steps S2 to S12 are repeated for the next reference image.

In step S7, the counting result is output and displayed.

Claims (4)

1. An audio skipping rope counting method based on a cross correlation coefficient method is characterized by comprising the following steps:

1) acquiring video data of rope skipping actions, and extracting audio data from the video data;

2) extracting the sampling frequency of the audio, and drawing an audio waveform diagram with time as an abscissa;

3) selecting sample sound, and calculating rope skipping times by adopting a cross-correlation coefficient method;

4) outputting and displaying a counting result;

in step 3), selecting sample sound, and calculating the rope skipping times by adopting a cross-correlation coefficient method, wherein the steps are as follows:

3-1) taking an audio fragment of a single impact between a skipping rope and the ground as a sample audio x, taking an audio sequence in a skipping rope video as a tested audio y, and calculating a cross-correlation coefficient between the sample audio x and the tested audio y;

3-2) taking a sound fragment of single impact of the skipping rope and the ground as sample audio; randomly taking the sound segments of the n-time rope skipping and the single impact on the ground for weighted average;

3-3) drawing a 'cross correlation coefficient' graph, setting a cross correlation coefficient threshold value, and counting the times m that the cross correlation coefficient exceeds the set threshold value, wherein m is the number of rope skipping;

in step 3-1), the cross-correlation function of the sample audio x and the tested audio y is:

wherein N is the length of the longer signal sequence in x, y, and the asterisks indicate complex conjugation; m represents the number of sampling points in translation;

representing the result of point-by-point corresponding multiplication of the two sequences after the sequence x is kept still and the sequence y is shifted to the left by m sampling points;

and (4) representing the result of point-by-point corresponding multiplication of the two sequences after the sequence y is kept still and the sequence x is shifted to the left by m sampling points.

2. The audio frequency skipping rope counting method according to claim 1, wherein in the step 3-2), n is 3.

3. The audio skipping rope counting method according to claim 1, wherein in step 3-3), the threshold value of the cross-correlation coefficient is set to 0.1.

4. The audio skipping rope counting method of claim 1, wherein at least 8820 data are set between two skipping rope counts.

Images