Living body motion judgment and feature extraction method based on ultrasonic echoes

Abstract

The invention discloses a living body motion judgment and feature extraction method based on ultrasonic echoes, and relates to the technical field of ultrasonic detection by using an MEMS sensor, and the method comprises the steps: segmenting an ultrasonic echo signal into N windows according to flight time, initializing an operation environment, and obtaining a detection result matrix of the N windows according to frames; and carrying out algorithm analysis on the result matrix, and judging whether a living body target approaches or is far away from the motion change or is air disturbance. According to the method, by extracting the motion characteristics of the living body in the ultrasonic echo signals, changes generated by the motion of the living body and changes caused by air disturbance in the environment are effectively distinguished, the accuracy of judging the existence of the motion of the living body is improved, and meanwhile, the overall trend slope of the stepped diagonal characteristics is obtained; the method can achieve the judgment of the approaching or departing direction of the living body.

Description

Technical Field

[0001] The present invention relates to the technical field of ultrasonic detection using MEMS sensors, and particularly to a method for judging the movement of a living body and extracting features based on ultrasonic echoes. Background Art

[0002] At present, there are many scenarios for ultrasonic algorithms. Among them, ultrasonic algorithms for judging the movement scenario of a living body target are prone to misjudgment due to the complexity of the environment and the uncertainty of the movement of the living body target. For the judgment of a moving living body target, the generally common ultrasonic algorithm is to simply judge whether there is a movement of the living body target by comparing the signal changes between multiple frames or between two adjacent frames. This algorithm can indeed effectively cancel the interference of static obstacle echoes. However, in an environment with a large influence of air disturbance, some intervals of the signal fluctuate up and down, so that it is impossible to distinguish whether these changes are caused by the movement of the living body target or by interference. In such a scenario, this algorithm cannot be used.

[0003] Therefore, aiming at the problem that the ultrasonic echo algorithm applied to the environment with a large influence of air disturbance is prone to misjudgment in judging the movement of a living body target.

[0004] For example, the invention application with the application number 202411070087.X discloses an obstacle recognition method, device, electronic device and storage medium. This solution greatly reduces the complexity of real-time calculation through a pre-established obstacle pattern library, improves the execution efficiency of the obstacle recognition algorithm, and dynamically adjusts the window size to reduce unnecessary calculation overhead while ensuring detection accuracy to meet the real-time requirements of the system. However, its solution cannot identify the misjudgment problem caused by air disturbance.

[0005] Therefore, in reality, there is a need for a method that can effectively distinguish the change characteristics generated by the movement of a living body target and the change characteristics caused by air disturbance in the environment by extracting the movement characteristics of the living body target in the ultrasonic echo signal, so as to improve the accuracy of judging the existence of the movement of the living body target. Summary of the Invention

[0006] Aiming at the above existing problems, the purpose of the present invention is to provide a method for judging the movement of a living body and extracting features based on ultrasonic echoes, which is used to identify the movement changes of a living body target approaching or moving away or air disturbance during the detection process.

[0007] The purpose of the present invention can be achieved by the following technical solutions: A method for judging the movement of a living body and extracting features based on ultrasonic echoes, comprising the steps of:

[0008] S1. Divide the ultrasonic echo signal into N windows according to the flight time, initialize the operating environment, and obtain the detection result matrix of the N windows frame by frame;

[0009] S2. Analyze the result matrix algorithmically to determine whether there is a living target approaching or moving away, or if it is an air disturbance.

[0010] As a further aspect of the present invention, when initializing the operating environment in S1, the ultrasonic echo signal sample data without a living target is stored in a reference array; the reference array is updated after the detection result matrix is obtained, and the current frame sample data is used to replace the reference array data as the new reference array, and the new reference array will be used to compare with the sample data of the next frame for judgment.

[0011] As a further aspect of the present invention, the steps for obtaining the result matrix in S1 include:

[0012] S11. Calculate the average difference between the current window samples and the corresponding window samples in the reference array, and perform difference integration on all samples in the current window that are greater than the average difference.

[0013] S12. Perform distance compensation on the difference integration. The farther the distance, the greater the compensation coefficient is multiplied by the window with a longer flight time.

[0014] S13. When the difference integration value of all samples in the window > the change threshold, the current window pane is set to 1; when the integration value of the differences of all samples ≤ the change threshold, the current window pane is set to 0.

[0015] S14. Repeat S11 - S13 to traverse all windows in turn to obtain the single-frame pane value.

[0016] S15. Repeat S11 - S14 for the next frame.

[0017] S16. Complete the traversal of all frames to form the result matrix.

[0018] As a further aspect of the present invention, in S2, when determining whether there is a living target approaching or moving away, the slope flag is initialized to 0, and by changing the slope flag, the slope distribution of the living target movement characteristics in the result matrix is classified and discussed. When:

[0019] The slope flag = 0, traverse the result matrix from near to far to determine whether there is a living target movement change. If there is no change, the slope flag + 1.

[0020] The slope flag = 1, traverse the result matrix from far to near to determine whether there is a living target movement change. If there is no change, the slope flag + 1.

[0021] The slope flag > 1, determine that there is no active movement.

[0022] As a further aspect of the present invention, the determination of whether there is a living target movement change includes the steps:

[0023] S21. Traverse the result matrix pane, assume and record the current window as the initial window;

[0024] S22. Record the first frame value with a pane value of 1 in the initial window. If there is no pane with a value of 1 in the initial window, traverse to the next window and repeat step S21;

[0025] S23. Traverse to the next window, and start traversing the panes from the same frame as the first frame with a pane value of 1 in the initial window for the current window's pane values

[0026] S24. When all the segments between the first frame with a pane value of 1 in the initial window and the segment with a pane value of 1 in the current window are segments with a value of 0, and the frame number (segment) of the first frame with a pane value of 1 in the current window is later than that of the first frame with a pane value of 1 in the previous non-empty window, the number of active windows +1; otherwise, traverse to the next window;

[0027] S25. Repeat steps S21 - S24. When the number of active windows ≥ 4, it is determined that there is a change in the movement of the living target.

[0028] As a further aspect of the present invention, the determination of air disturbance in S2 includes:

[0029] When the segment with a pane value of 1 is before the first frame segment with a pane value of 1 in the initial window, or the segment with the current pane value of 1 is more forward than the first frame segment with a pane value of 1 in the previous traversed non-empty window, it is determined that the segment is an empty window segment caused by air disturbance.

[0030] As a further aspect of the present invention, in S25, when the number of active windows ≥ 4, the approach or departure of the living target is determined according to the change trend of the frame values of the active windows.

[0031] Advantages of the present invention:

[0032] 1. The present invention effectively distinguishes the changes generated by the movement of the living body and the changes caused by air disturbance in the environment by extracting the movement characteristics of the living body in the ultrasonic echo signal, improving the accuracy of determining the existence of living body movement.

[0033] 2. The present invention can determine the direction of approach or departure of the living body movement by obtaining the overall trend slope of the stepped diagonal feature. Brief Description of the Drawings

[0034] Figure 1 It is a schematic diagram of the principle structure for judging the movement of the living target in the present invention;

[0035] Figure 2 It is a schematic diagram of the principle structure for judging air disturbance in the present invention;

[0036] Figure 3 This is a flowchart of the method for judging the movement of a living target in the present invention. Detailed implementation manners

[0037] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar symbols represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation of the present invention.

[0038] When using ultrasonic detection to determine the movement of a living target, due to the influence of air disturbance in the scene environment and the uncertainty of the movement of the living target, false judgments are likely to occur.

[0039] To address the above problems, as Figure 3 shown in the flowchart, the present invention discloses a method for judging the movement of a living body and extracting features based on ultrasonic echoes, including the steps of:

[0040] S1. Divide the ultrasonic echo signal into N windows according to the flight time, initialize the operating environment, and obtain the detection result matrix of N windows frame by frame;

[0041] S2. Perform algorithm analysis on the result matrix to determine whether there is a change in the movement of a living target approaching or moving away.

[0042] As Figure 1 shown, divide the detection area into N windows according to the flight time of the ultrasonic echo signal, and each window corresponds to a block of the signal area.

[0043] Divide the signal frames according to the detection cycle time. For example, one cycle is divided into M frames, and a matrix is formed by N windows and M frames.

[0044] First, initialize the operating environment and store the ultrasonic echo signal sample data without a living target in the reference array.

[0045] First, calculate the average difference between the sample points of the current window and the corresponding window sample points in the reference array, perform difference integration on all sample points in the current window that are greater than the average difference, and obtain the integral value of each window. The integral value reflects the change amount of the current frame of each window relative to the previous reference frame. The more the change, the larger the integral value; integrating all sample points > average difference within each window is to retain the large change features and eliminate the smaller interferences to reduce the interference of errors such as electrical signal jitter.

[0046] Since the detection distance is farther, the signal attenuation is more. When integrating all the sample points within the window, the integration value is compensated according to the distance between the window and the ultrasonic detector. For each window, the integration is multiplied by a different compensation coefficient according to the distance. The farther the distance, the more the compensation. Then, a fixed threshold is used to determine whether each window has changed.

[0047] Then, based on the integration value, it is determined whether there is a change within the window, and a result matrix is obtained. For example: The first frame initializes the operating environment. The sample data of each window in the first frame is used as a reference array, and the difference between the calculated sample point integration value and the window reference array is calculated. When the difference > the change threshold, the pane of window No. 1 is set to 1. When the difference ≤ the change threshold, the pane of window No. 1 is set to 0. Repeat the above steps to traverse all windows in turn.

[0048] Using the above traversal method, subsequent frame calculations are performed until all frames are traversed, and a result matrix is formed based on all frames and all windows.

[0049] Utilizing the characteristic that the farther the ultrasonic echo target is, the longer the flight time. When a living target approaches from a distance, the first window to change is the one with the longest flight time where the echo of the living target is located. Then this change will slowly move towards the window with a shorter flight time, corresponding to the approach in distance in space.

[0050] Since the frame rate of the detection signal is generally much greater than the movement speed of the living target, the result matrix can well record the changes generated in each frame when the living target moves. Form as Figure 1 the result matrix pane shown.

[0051] The law of the echo signal changes caused by the movement of the living target from far to near in different frames of each window is manifested as a gradually decreasing stepped diagonal line. This stepped diagonal line is the movement feature of the living target extracted after a series of processes such as segmentation, difference calculation, and integration.

[0052] With this movement feature, even in the presence of air disturbances, it is possible to distinguish the changes caused by air disturbances and the changes generated by the movement of the living target in the seemingly chaotic and changing echo signal as a whole.

[0053] Furthermore, according to the overall trend slope of this stepped diagonal line feature, it can be determined whether the living target is approaching (slope > 0) or moving away (slope < 0).

[0054] Based on the above principle, the result matrix is analyzed, as shown in Figure 1 and Figure 2 shown.

[0055] Set a slope flag. When determining whether there is a change in the movement of a living target approaching or moving away, first initialize the slope flag to 0. By changing the slope flag, classify and discuss the slope distribution of the living movement characteristics in the judgment result matrix. When:

[0056] When the slope flag = 0, traverse the result matrix from near to far to determine whether there is a change in the movement of the living target. If there is no change, increment the slope flag by 1;

[0057] When the slope flag = 1, traverse the result matrix from far to near to determine whether there is a change in the movement of the living target. If there is no change, increment the slope flag by 1;

[0058] When the slope flag > 1, determine that there is no active movement and end the analysis of the detection result matrix for one cycle.

[0059] Take the case when the slope flag = 1 as an example. At this time, traverse the result matrix from far to near to determine whether there is a change in the movement of the living target.

[0060] According to the change law of the result matrix composed of all windows, check whether there is a change characteristic of a stepped diagonal line among them. If there is such a characteristic and the length of the stepped shape meets a certain number of windows, it indicates that the change is caused by the movement of the living target passing by; if there is a disconnection during the stepped period, it means that the living body moves relatively fast, but generally there will be no disconnection of more than 2 windows.

[0061] The windows forming the steps are active windows. When the number of active windows ≥ 4, determine whether the living target is approaching or moving away based on the change trend of the frame values of the active windows.

[0062] As Figure 1 shown, take an example: Traverse the panes of the result matrix. When the value of a pane is 1, set the current window as the initial window. For example, when detecting the situation of a living body approaching (traversing from back to front), there is no pane value of 1 in windows 1 to 5, and the pane value of the first frame of window 6 is 1. Therefore, record window 6 as the initial window.

[0063] Continue to traverse the subsequent windows, starting from the same frame where the first pane value of the initial window is 1. For example, since the pane value of the first frame of the 6th window is 1, subsequent window traversals all start from the same frame as the initial window, that is, the first frame, to traverse the grid segments. The first pane value of the 7th window that is 1 is also the first frame, which does not meet the condition that the first frame with a pane value of 1 is later than the previous window. So, skip the 7th window and traverse to the 8th window. The first pane value of the 8th window that is 1 is the second frame. At the same time, it meets the two conditions: "All grid segments between the first frame with a pane value of 1 in the initial window and the grid segment with a pane value of 1 in the current window are grid segments with a value of 0" and "The frame number (grid segment) of the first frame with a pane value of 1 in the current window is later than the first frame with a pane value of 1 in the previous non-empty window", that is: "The first frame of the 8th window is 0, and the second frame is 1", and "The second frame of the 8th window is later than the first frame of the 7th window". Therefore, it is determined that there are characteristics of a living body passing through the 8th window, and the number of windows passed by the living body movement trajectory is incremented by 1. Similarly, since the 9th window has the first segment with a value of 1 in the same frame as the 8th window, it is directly skipped and traversed to the 10th window. The 10th, 11th, and 12th windows all meet the two characteristic rules of living body movement: "All grid segments between the first frame with a pane value of 1 in the initial window and the grid segment with a pane value of 1 in the current window are grid segments with a value of 0" and "The frame number (grid segment) of the first frame with a pane value of 1 in the current window is later than the first frame with a pane value of 1 in the previous non-empty window". The number of windows passed by the living body movement trajectory reaches 4. Therefore, it is determined that there is a moving living body, and at this time, it is traversing from far to near, and its distribution slope is positive. Therefore, it is determined to be a approaching action.

[0064] In the above algorithm for calculating the number of active windows, finding the feature of the number of active windows is actually achieved by checking whether the non-changing frames on the first changing frame of each window are normal. The reason for this judgment is that in the flight time period corresponding to the area where the living body has not arrived, there will definitely be no premature changes, while in the flight time period corresponding to the area where the living body passes through, due to the influence of multiple uncertain echo reflection paths, it may not immediately end the change or always maintain a changing state. If directly judging the windows with changes, the algorithm will become more complex, while judging the non-changing windows is more stable and accurate.

[0065] Furthermore, as Figure 2 shown, when there is air disturbance, the matrix values obtained by the air disturbance disrupt the step slope formed by the result matrix of the living body target, forming non-adjacent grid segments. Based on this feature, it can be distinguished whether the change is caused by air disturbance or living body movement.

[0066] As Figure 2 shown, an example is given: When traversing the grid pane of the result matrix, when there is a grid pane value of 1, set the current window as the initial window. For example, the grid pane value of the second frame of the 4th window is 1, and record the 4th window as the initial window.

[0067] Since there are no panes with a value of 1 in the 5th and 6th windows, they are both empty windows. When the number of empty windows is greater than or equal to 2 in a row, it is determined that the 4th window segment in front of the empty window is caused by air disturbance. At this time, the empty window is skipped, the initial window is re-determined, and the traversal continues. The 7th and 8th windows meet the characteristic law of living body movement, and the 7th window is the initial window. Although the 9th and 10th windows are not empty windows, they do not meet the two conditions of "the first frame with a pane value of 1 in the initial window to the segment with a pane value of 1 in the current window are all 0 segments" and "the number of frames (segments) of the first frame with a pane value of 1 in the current window is later than the first frame with a pane value of 1 in the previous non-empty window". Therefore, they do not belong to the regular distribution characteristics of living body movement. It is determined that these two windows are mixed with the interference of air disturbance, so they are not counted as living body movement characteristics and are directly skipped. The 11th, 12th, and 13th windows meet the characteristics of living body motion. So far, starting from the initial window 7, the 8th, 11th, 12th, and 13th, there are more than 4 windows that meet the diagonal distribution, and there are no empty windows in between, so the living body motion characteristics are extracted from the matrix mixed with interference.

[0068] The present invention divides the ultrasonic echo signal into N windows according to the flight time, each window corresponds to a region of the signal and also corresponds to a column in the result matrix.

[0069] Each window integrates the sample data with a value greater than the average difference in the window, thereby obtaining the difference integral value of each window. This value reflects the amount of change of the current frame of each window relative to the previous reference frame. The more changes, the greater the integral value. Among them, the integral term is greater than the average difference in order to retain the large change features and eliminate the smaller interference to reduce the interference of errors such as electrical signal jitter. Then, according to the size of this integral value, it is judged whether there is a change in the window. If there is a change, it is set to 1, and if there is no change, it is cleared to 0. The change results of each frame of each window are stored in the result matrix. Taking advantage of the characteristic that the farther the ultrasonic echo target is, the longer the flight time is. When the living body approaches from a distance, the first change is the window with the longest flight time where the moving living body echo is located, and then this change will slowly move to the window with a short flight time, corresponding to the distance approaching in space. Since the detection frame rate is generally much greater than the movement speed of the living body, the result matrix can well record the changes in each frame when the living body moves. The movement of living things from far to near causes the change of echo signals in different frame numbers of each window as a gradually descending step-shaped diagonal line. This step-shaped diagonal line is the motion feature of living things extracted after a series of segmentation, difference, integration and other processing. With this motion feature, even if there is air disturbance, the changes caused by air disturbance and the changes caused by the movement of living things can be distinguished in the entire seemingly chaotic echo signal.

[0070] The above are only the preferred specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, shall be covered by the protection scope of the present invention.

[0071] In the present invention, unless otherwise clearly defined and limited, the first feature being "on" or "under" the second feature may be that the first and second features are in direct contact, or the first and second features are indirectly in contact through an intermediate medium. Moreover, the first feature being "above", "over" and "on top of" the second feature may be that the first feature is directly above or obliquely above the second feature, or merely indicates that the first feature has a higher level height than the second feature. The first feature being "under", "below" and "beneath" the second feature may be that the first feature is directly below or obliquely below the second feature, or merely indicates that the first feature has a lower level height than the second feature.

Claims

1. A method for judging living body movement and feature extraction based on ultrasonic echo, characterized in that Including: S1. Divide the ultrasonic echo signal into N windows according to the flight time, initialize the operating environment, and obtain the detection result matrix of N windows frame by frame; S2. Perform algorithm analysis on the result matrix to determine whether there is a living target approaching or moving away or air disturbance.

2. The method for judging and extracting features of living body movement based on ultrasonic echo according to claim 1, wherein When initializing the operating environment in S1, store the ultrasonic echo signal sample data without a living target into the reference array; the reference array is updated after the detection result matrix is obtained, and the current frame sample data is used to replace the reference array data as the new reference array, and the new reference array will be used to compare and judge with the sample data of the next frame.

3. A method for judging and extracting features of living body movement based on ultrasonic echo according to claim 1, characterized in that, In S1, the steps for obtaining the result matrix include: S11. Calculate the average difference between the current window sample points and the corresponding window sample points of the reference array, and perform difference integration on all sample points in the current window that are greater than the average difference; S12. Perform distance compensation on the difference integration, and the window with a longer flight time due to a farther distance is multiplied by a larger compensation coefficient; S13. When the integral value of the difference of all sample points in the window > the change threshold, set the current window pane to 1, and when the integral value of the difference of all sample points ≤ the change threshold, set the current window pane to 0; S14. Repeat S11 - S13 to traverse all windows in turn to obtain the single-frame pane value; S15. Repeat S11 - S14 for the next frame; S16. Complete the traversal of all frames to form the result matrix.

4. A method for judging and extracting features of living body movement based on ultrasonic echo according to claim 1, characterized in that In S2, when judging whether there is a living target approaching or moving away, initialize the slope flag to 0, and by changing the slope flag, classify and discuss the slope distribution of the living target motion characteristics in the result matrix. When: The slope flag = 0, traverse the result matrix from near to far, judge whether there is a living target motion change, and if there is no change, the slope flag + 1; The slope flag = 1, traverse the result matrix from far to near, judge whether there is a living target motion change, and if there is no change, the slope flag + 1; When the slope flag > 1, judge that there is no active motion.

5. A method for judging and extracting features of living body movement based on ultrasonic echo according to claim 4, characterized in that, The judgment of whether there is a living target motion change includes the steps: S21. Traverse the result matrix pane, assume and record the current window as the initial window; S22. Record the value of the first frame when the pane value of the initial window is 1. If there is no pane value of 1 in the initial window, traverse to the next window and repeat step S21; S23. Traverse to the next window, and start traversing the pane from the same frame as the first frame when the pane value of the initial window is 1; S24. When the segment between the first frame when the pane value of the initial window is 1 and the first frame when the pane value of the current window is 1 is all segments with a value of 0, and the number of frames (segments) of the first frame when the pane value of the current window is 1 is later than that of the first frame when the pane value of the previous non-empty window is 1, the number of active windows + 1, otherwise traverse to the next window; S25. Repeat steps S21 - S24. When the number of active windows ≥ 4, judge that there is a living target motion change.

6. A method for judging and extracting features of living body movement based on ultrasonic echo according to claim 5, characterized in that, In S2, the judgment of air disturbance includes: When the segment with a pane value of 1 is before the first frame segment with an initial window pane value of 1, or when the segment with the current pane value of 1 is more forward than the first frame segment with a non-empty window pane value of 1 in the previous traversed window, the segment is determined to be an empty window segment caused by air disturbance.

7. A method for judging and extracting features of living body movement based on ultrasonic echo according to claim 6, characterized in that, In S25, when the number of active windows ≥ 4, the approach or departure of the living target is determined based on the change trend of the active window frame value.

Images