Intelligent traffic police mobile video monitoring method and system

Abstract

The invention discloses an intelligent traffic police mobile video monitoring method and system, relates to the technical field of video monitoring, and solves the problems that a processing mode is not accurate enough, a corresponding video still shakes too much, a determined monitoring video is not high enough in temperature, and the overall effect of the monitoring video cannot be improved. According to the method, when the associated feature points in the next frame of picture are determined, the pixel sequences associated with the corresponding pixel points are determined, the determined multiple groups of pixel sequences and the pixel sequences of the associated feature points are verified and analyzed, the associated gradient sequence difference values are determined, and in the verification process based on the multiple groups of gradient sequence difference values, the pixel sequences associated with the corresponding pixel points in the next frame of picture are verified. And the feature points are selected, so that the determination of the feature points is more accurate, the real-time output of the correlation vector is ensured, the correlation accuracy of the movement of the monitoring probe is ensured, and a better video monitoring effect is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of video surveillance, and in particular to a smart traffic police mobile video surveillance method and system. Background Art

[0002] In the field of traffic management, smart traffic police mobile video surveillance is becoming a key technical means to improve the level of traffic management and ensure road safety; with the rapid development of science and technology, traditional traffic monitoring methods have gradually revealed their limitations, and the smart traffic police mobile video surveillance system that integrates advanced technology has brought a new solution to traffic management.

[0003] The application with publication number CN103595965A discloses a mobile video monitoring method based on video flow control, including a process in which a mobile terminal obtains monitoring video through a mobile communication network, the process including the following steps: reading and decoding the original video stream, converting it into an image sequence; performing flow regulation on the image sequence to obtain a flow-regulated image sequence; encoding the flow-regulated image sequence, converting it into a transmission video stream with a video format compatible with the mobile terminal; transmitting the transmission video stream to the mobile terminal through a mobile communication network. The present invention reduces video flow by pre-controlling the monitoring video flow before transmitting it through a mobile communication network, thereby reducing the bandwidth occupied by video monitoring and the flow cost generated.

[0004] During the mobile video monitoring process, feature objects are generally selected based on the relevant features of the monitoring screen, and the feature objects are used as standards to control the relevant monitoring equipment in real time to ensure the relevant stability of the monitoring video. However, this method is not accurate enough, which will cause the corresponding video to still have excessive jitter. The determined monitoring video is not accurate enough and cannot improve the overall effect of the monitoring video. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a smart traffic police mobile video monitoring method and system, which solves the problem that the processing method is not accurate enough, resulting in excessive jitter in the corresponding video, and the determined monitoring video is not warm enough, and the overall effect of the monitoring video cannot be improved.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a smart traffic police mobile video monitoring method, comprising the following steps:

[0007] Step 1: Based on the total coverage area of ​​the images monitored by the monitoring equipment, a group of central areas are determined from the total coverage area, and the images associated with the central areas are marked as central images. The monitoring images of different frames are all confirmed to have associated central images. The specific sub-steps are as follows:

[0008] S11, based on the display picture determined during the monitoring process of the monitoring device, confirming the center point of the display picture, wherein the display picture is displayed by a related display device, and the center point is preset in the display device;

[0009] S12, based on the center point of the display picture and the specific outline of the display picture, the specific outline of the display picture is reduced by X1 times, where X1 is a preset value, and the picture associated with the corresponding reduced area after the reduction is marked as the center picture of the current frame picture;

[0010] Step 2: Perform feature point analysis on the confirmed central image in the current frame, confirm the pixel values ​​of different pixels in the central image, and then confirm the gradient features associated with the corresponding pixels. Based on the different gradient features associated with different pixels, feature points are determined from different pixels. The specific sub-steps are as follows:

[0011] S21, based on the central picture determined in the current frame, confirm the pixel values ​​associated with different pixel points in the central picture, and mark them as X i , where i represents different pixels. Based on other pixels around the corresponding pixel, the Sobel algorithm is used to confirm the horizontal and vertical gradients of the pixel, and the horizontal gradient of the pixel is calibrated as H i , the vertical gradient is calibrated as S i ;

[0012] S22, based on the lateral gradient H confirmed by the corresponding pixel point i And the vertical gradient S i , confirm the comprehensive gradient ZH of this pixel i ,in Based on the comprehensive gradient confirmed by this pixel point, the four groups of pixels associated with the upper and lower positions and the left and right positions of this pixel point are confirmed to generate a set of pixel sequences: Among them SZ i and XZ i Represents the pixel point associated with the upper position and the lower position, whose ZX i and YX i Represents the pixel points associated with the left and right positions, and SZ i , X i and XZ i The three groups of comprehensive gradients are averaged to confirm the first average gradient T1 i , and then ZX i , X i and YX i The three groups of comprehensive gradients are averaged to confirm the second average gradient T2 i ;

[0013] Based on the confirmed first mean gradient T1 i and the second mean gradient T2 i , lock T1 i and T2 i The lowest common multiple of , and the locked common multiple is used as the gradient feature of this pixel;

[0014] S23, based on different gradient features confirmed by different pixel points, a maximum value is selected from the confirmed groups of gradient features, and the pixel point corresponding to the maximum value is used as a feature point associated with the current frame;

[0015] Step 3: After the next set of frames located at the current frame is generated, the next set of frames is processed in the same way as step 2, the center frame is confirmed first, and the gradient features associated with different pixels in the center frame are confirmed, the feature points confirmed by the previous frame are locked, and the associated vector is confirmed based on the position change associated with the feature points. The specific sub-steps are:

[0016] S31, the next frame of the generated picture is processed in the same manner as in step 2, the central picture is preferentially determined, and then the gradient features associated with each different pixel point in the central picture are confirmed in the same manner as in steps S21-S22;

[0017] S32, the gradient features associated with the feature points in the previous frame are recorded as upper gradient features, and the gradient features associated with different pixel points in the next frame are recorded as lower gradient features:

[0018] From the several groups of lower gradient features confirmed in the next frame, identify the pixel points with the same lower gradient features as the upper gradient features. If there are related pixel points, record the existing pixel points as the moving points of the feature points. Take the position of the feature points as the starting point and the position of the moving points as the end point to confirm a group of correlation vectors. The confirmed moving points are the feature points of this frame, which are used for the confirmation of the correlation vectors of the next frame.

[0019] If there are no relevant pixel points, the pixel sequence associated with the feature point is confirmed and recorded as the main pixel sequence, and based on the location of the feature point, the same position point is confirmed in the next frame, and the confirmed position point is used as the center of the circle, and then a group of radial circles is confirmed with a radius value of R1, where R1 is a preset value, and the pixel sequences associated with different pixel points in the radial circle are confirmed one by one, and the confirmed pixel sequences are recorded as sub-pixel sequences, and the gradient features at the same sequence position of the main pixel sequence and the sub-pixel sequence are difference processed to confirm the gradient sequence difference at the same sequence position. The difference processing method of each same sequence position is the same, and several groups of gradient sequence difference values ​​belonging to the same sub-pixel sequence are proofread to identify whether the several groups of gradient sequence difference values ​​are consistent. If they are consistent, the pixel point associated with this pixel sequence is marked as a moving point. If they are not consistent, the gradient sequence difference values ​​associated with other sub-pixel sequences are continuously confirmed and proofread until the moving point is confirmed. If the moving point has not been confirmed, the feature point of the current frame is re-confirmed in the same way as step 2;

[0020] Taking the position of the feature point as the starting point and the position of the moving point as the end point, a set of associated vectors is confirmed. The confirmed moving point is the feature point of the current frame, which is used for the confirmation of the associated vector of the next frame.

[0021] Step 4: Based on the correlation vectors confirmed between adjacent frames, the monitoring device is controlled in real time to move the monitoring probe inside the monitoring device in the opposite direction. The specific method is as follows:

[0022] The monitoring probe is controlled to move in the opposite direction in real time, with the end point of the correlation vector as the starting point and the starting point of the correlation vector as the end point. The monitoring probe is controlled in real time based on the correlation vectors confirmed between different adjacent frames.

[0023] Preferably, a smart traffic police mobile video surveillance system comprises:

[0024] The center picture calibration end determines a group of center areas from the total coverage area based on the total coverage area of ​​the pictures monitored by the monitoring equipment, and calibrates the pictures associated with the center areas as the center pictures. The monitoring pictures of different frames are all confirmed to have associated center pictures.

[0025] The feature point calibration end performs feature point analysis on the central image confirmed in the current frame, confirms the pixel values ​​of different pixels in the central image, and then confirms the gradient features associated with the corresponding pixels. Based on the different gradient features associated with different pixels, feature points are determined from different pixels.

[0026] The frame vector confirmation end is located at the next group of frame images of the current frame image, and processes the next group of frame images in the same manner as step 2, giving priority to confirming the central image, and confirming the gradient features associated with different pixel points in the central image, locking the feature points confirmed by the previous frame image, and confirming the associated vector based on the position change associated with the feature points;

[0027] The probe control adjustment end controls the monitoring device in real time based on the correlation vector confirmed between adjacent frame images, so that the monitoring probe inside the monitoring device moves in the opposite direction.

[0028] The present invention provides a smart traffic police mobile video monitoring method and system. Compared with the prior art, it has the following beneficial effects:

[0029] The present invention determines the pixel sequence associated with the corresponding pixel point when determining the associated feature point in the next frame, verifies and analyzes the confirmed pixel sequences and the pixel sequence of the associated feature point, and determines the associated gradient sequence difference. Based on the verification process of the gradient sequence difference, the feature point is selected, which can make the feature point determination more accurate, thereby ensuring the real-time output of the associated vector, ensuring the relevant accuracy of the monitoring probe movement, and achieving a better video monitoring effect.

[0030] Accurately determine feature points from the center picture pixel points; this not only can keenly capture the changes in key information in the picture, but also provides a reliable basis for the subsequent calculation of correlation vectors. In this way, the moving direction of the video shooting can be quickly identified, and timely response and adjustment can be made, effectively reducing the picture blur or offset caused by the movement of the monitoring equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the process of the present invention;

[0032] Figure 2 It is a schematic diagram of the principle framework of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] First embodiment

[0035] See also Figure 1 , the present application provides a smart traffic police mobile video monitoring method, comprising the following steps:

[0036] Step 1: Based on the total coverage area of ​​the images monitored by the monitoring equipment, a group of central areas are determined from the total coverage area, and the images associated with the central areas are calibrated as the central images. The monitoring images of different frames are confirmed to have associated central images. Specifically, the monitored video is composed of several different frames, that is, during monitoring, the central images associated with adjacent frames will shake slightly. In this case, there are related jitter parameters in the jitter process. By confirming such parameters later, the effect of video anti-shake can be effectively achieved. The specific sub-steps for calibrating the central image are as follows:

[0037] S11, based on the display picture determined during the monitoring process of the monitoring device (that is, the relevant picture displayed on the display terminal after monitoring), confirming the center point of the display picture, wherein the display picture is displayed by the relevant display device, and the center point is preset in the display device;

[0038] S12, based on the center point of the display picture and the specific outline of the display picture, the specific outline of the display picture is reduced by X1 times, where X1 is a preset value, generally 0.3, which is prepared in advance by relevant operators based on experience, and the picture associated with the corresponding reduced area after the reduction is marked as the center picture of the current frame picture;

[0039] Specifically, when the picture is being monitored, it will be displayed through the corresponding display terminal. Then, in the displayed picture, there will be a relevant central area. Based on the determined central point and the way of reducing the outline of the overall picture, the central picture in the middle position can be locked, so that the feature points can be confirmed in the locked central picture. Based on the movement mode of the same feature points between adjacent frames, the specific spatial vector of the movement is determined, and timely adjustment is made for anti-shake, so as to achieve a better mobile video monitoring effect.

[0040] Step 2: Perform feature point analysis on the confirmed central picture in the current frame, confirm the pixel values ​​of different pixels in the central picture, and then confirm the gradient features associated with the corresponding pixels. Based on the different gradient features associated with different pixels, feature points are determined from different pixels. Specifically, the gradient features associated with the pixels include horizontal gradient features and vertical gradient features. The gradient features are comprehensive features of horizontal gradients and vertical gradients, which can fully demonstrate the gradient feature of the pixels. Based on the specific performance of the gradient features, a group of pixels with the strongest gradient feature performance are selected as relevant feature points.

[0041] Among them, the specific sub-steps of determining feature points from different pixel points are:

[0042] S21, based on the central picture determined in the current frame, confirm the pixel values ​​associated with different pixel points in the central picture, and mark them as X i , where i represents different pixels. Based on other pixels around the corresponding pixel, the Sobel algorithm is used to confirm the horizontal and vertical gradients of the pixel, and the horizontal gradient of the pixel is calibrated as H i , the vertical gradient is calibrated as S i Specifically, the Sobel algorithm is used to confirm the gradient value of the pixel point by first combining the preset horizontal convolution kernel to confirm the horizontal gradient H i , and its horizontal convolution kernel is The pixel values ​​of the neighboring pixels around the pixel are calibrated as Z1-Z8, then after sorting, it is Based on the confirmed convolution kernel and the sorted pixel value sequence, the lateral gradient H is calculated. i Confirmation of H i =Z1×(-1)+Z2×0+Z3×1+Z4×(-2)+X i ×0+Z5×2+Z6×(-1)+Z7×0+Z8×1;

[0043] The convolution kernel associated with its vertical gradient is: The sequence of its pixel points after sorting is still Then after processing, the confirmed vertical gradient S i =Z1×(-1)+Z2×(-2)+Z3×1+Z4×0+X i ×0+Z5×0+Z6×1+Z7×2+Z8×1, where the convolution kernels are all fixed values;

[0044] S22, based on the lateral gradient H confirmed by the corresponding pixel point i And the vertical gradient S i , confirm the comprehensive gradient ZH of this pixel i ,in Based on the comprehensive gradient confirmed by this pixel point, the four groups of pixels associated with the upper and lower positions and the left and right positions of this pixel point are confirmed to generate a set of pixel sequences: Among them SZ i and XZ i Represents the pixel point associated with the upper position and the lower position, whose ZX i and YX i Represents the pixel points associated with the left and right positions, and SZ i , X i and XZ i The three groups of comprehensive gradients are averaged to confirm the first average gradient T1 i , and then ZXi , X i and YX i The three groups of comprehensive gradients are averaged to confirm the second average gradient T2 i ;

[0045] Based on the confirmed first mean gradient T1 i and the second mean gradient T2 i , lock T1 i and T2 i The lowest common multiple of , and the locked common multiple is used as the gradient feature of this pixel;

[0046] S23, based on different gradient features confirmed by different pixel points, a maximum value is selected from the confirmed groups of gradient features, and the pixel point corresponding to the maximum value is used as a feature point associated with the current frame;

[0047] Specifically, a set of feature points needs to be confirmed for each frame of different monitoring images. With the specific changes of the feature points, the shooting and moving direction of the corresponding video can be identified, so as to perform anti-shake processing on the monitored video in real time to improve the overall monitoring effect of the video monitoring and achieve better video processing effect;

[0048] Step 3: After the next set of frames located at the current frame is generated, the next set of frames is processed in the same manner as step 2, the center frame is confirmed first, and the gradient features associated with different pixel points in the center frame are confirmed, the feature points confirmed by the previous frame are locked, and the associated vector is confirmed based on the position change associated with the feature points, wherein the specific sub-steps of confirming the associated vector are:

[0049] S31, the next frame of the generated picture is processed in the same manner as in step 2, the central picture is preferentially determined, and then the gradient features associated with each different pixel point in the central picture are confirmed in the same manner as in steps S21-S22;

[0050] S32, the gradient features associated with the feature points in the previous frame are recorded as upper gradient features, and the gradient features associated with different pixel points in the next frame are recorded as lower gradient features:

[0051] From the several groups of lower gradient features confirmed in the next frame, identify the pixel points with the same lower gradient features as the upper gradient features. If there are related pixel points, record the existing pixel points as the moving points of the feature points. Take the position of the feature points as the starting point and the position of the moving points as the end point to confirm a group of correlation vectors. The confirmed moving points are the feature points of this frame, which are used for the confirmation of the correlation vectors of the next frame.

[0052] If there are no related pixel points, the pixel sequence associated with the feature point is confirmed and recorded as the main pixel sequence, and based on the location of the feature point, the same position point is confirmed in the next frame, and the confirmed position point is used as the center of the circle, and then a group of radial circles are confirmed with a radius value of R1, where R1 is a preset value, and its specific value is determined by the operator based on experience. The pixel sequences associated with different pixel points in the radial circle are confirmed one by one, and the confirmed pixel sequences are recorded as sub-pixel sequences. The gradient features of the same sequence position of the main pixel sequence and the sub-pixel sequence are difference processed, and the gradient sequence difference at the same sequence position is confirmed. The difference processing method for each same sequence position is the same (for example: the gradient features of the same sequence position of the main pixel sequence and the sub-pixel sequence are T1 and T2, and T1 is the main pixel The gradient features of the sequence, T2 is the gradient features of the sub-pixel sequence, and the difference value = T1-T2, so the differences at other positions in the same sequence are also processed in this way), and several groups of gradient sequence differences belonging to the same sub-pixel sequence are proofread to identify whether the gradient sequence differences are consistent. If they are consistent, the pixel point associated with this pixel sequence is marked as a moving point. If they are not consistent, the gradient sequence differences associated with other sub-pixel sequences are continuously confirmed and proofread until the moving point is confirmed. If the moving point is not confirmed, the feature points of the current frame are reconfirmed in the same way as step 2 (indicating that the center picture has been changed, and the vehicle may be in a turning path or other paths, causing its original feature points to deviate from the center picture, resulting in the inability to confirm the feature points);

[0053] Taking the position of the feature point as the starting point and the position of the moving point as the end point, a set of associated vectors is confirmed. The confirmed moving point is the feature point of the current frame, which is used for the confirmation of the associated vector of the next frame.

[0054] Specifically, after the associated feature points are confirmed in the previous frame, there is no corresponding feature point in the next frame, which means that there is a difference between the front and back of the corresponding frames, that is, the vehicle moves forward. For the same point, the related pixel values ​​associated with the same point will change, and the associated gradient features will also change. Therefore, the difference confirmation method can be used to confirm the same point and conduct a comprehensive assessment on the confirmed differences. Because the pixel values ​​of the corresponding pixels change uniformly when the vehicle is moving, the generated differences all belong to the same type of differences. Therefore, based on the confirmed same type of differences, the same feature points can be locked to confirm the associated moving points, and thus the corresponding associated vectors can be confirmed.

[0055] Step 4: Based on the correlation vector confirmed between adjacent frames, the monitoring device is controlled in real time to make the monitoring probe inside the monitoring device move in the opposite direction. The specific method of the control is as follows:

[0056] Taking the end point of the correlation vector as the starting point and the starting point of the correlation vector as the end point, the monitoring probe is controlled to move in the opposite direction in real time. Based on the correlation vector confirmed between different adjacent frames, the monitoring probe is controlled in real time to improve the monitoring efficiency of the monitoring equipment and the overall quality of the monitoring video.

[0057] Second embodiment

[0058] Combination Figure 2 , a smart traffic police mobile video monitoring system, comprising:

[0059] The center picture calibration end determines a group of center areas from the total coverage area based on the total coverage area of ​​the pictures monitored by the monitoring equipment, and calibrates the pictures associated with the center areas as the center pictures. The monitoring pictures of different frames are all confirmed to have associated center pictures.

[0060] The feature point calibration end performs feature point analysis on the central image confirmed in the current frame, confirms the pixel values ​​of different pixels in the central image, and then confirms the gradient features associated with the corresponding pixels. Based on the different gradient features associated with different pixels, feature points are determined from different pixels.

[0061] The frame vector confirmation end is located at the next group of frame images of the current frame image, and processes the next group of frame images in the same manner as step 2, giving priority to confirming the central image, and confirming the gradient features associated with different pixel points in the central image, locking the feature points confirmed by the previous frame image, and confirming the associated vector based on the position change associated with the feature points;

[0062] The probe control adjustment end controls the monitoring device in real time based on the correlation vector confirmed between adjacent frame images, so that the monitoring probe inside the monitoring device moves in the opposite direction.

[0063] Third embodiment

[0064] The specific implementation process of this embodiment includes the entire implementation process of the above two groups of embodiments.

[0065] Some of the data in the above formulas are dimensionless and numerically calculated. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0066] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A smart traffic police mobile video monitoring method, characterized in that: The following steps are involved: Step 1: Based on the total coverage area of ​​the images monitored by the monitoring device, a group of central areas are determined from the total coverage area, and the images associated with the central areas are marked as central images, and the monitoring images of different frames are all confirmed to have associated central images; Step 2: Perform feature point analysis on the confirmed central picture in the current frame, confirm the pixel values ​​of different pixels in the central picture, and then confirm the gradient features associated with the corresponding pixels. Based on the different gradient features associated with different pixels, feature points are determined from different pixels. Step 3: After the next set of frames located at the current frame is generated, the next set of frames is processed in the same manner as step 2, the center frame is confirmed first, and the gradient features associated with different pixel points in the center frame are confirmed, the feature points confirmed by the previous frame are locked, and the associated vector is confirmed based on the position change associated with the feature points; Step 4: Based on the correlation vector confirmed between adjacent frames, the monitoring device is controlled in real time to make the monitoring probe inside the monitoring device move in the opposite direction.

2. The intelligent traffic police mobile video monitoring method according to claim 1 is characterized in that: In step 1, the specific sub-steps for calibrating the center picture are: S11, based on the display picture determined during the monitoring process of the monitoring device, confirming the center point of the display picture, wherein the display picture is displayed by a related display device, and the center point is preset in the display device; S12, based on the center point of the display picture and the specific outline of the display picture, the specific outline of the display picture is reduced by X1 times, wherein X1 is a preset value, and the picture associated with the corresponding reduced area after the reduction is marked as the center picture of the current frame picture.

3. The intelligent traffic police mobile video monitoring method according to claim 1 is characterized in that: In step 2, the specific sub-steps of determining the feature points are: S21, based on the central picture determined in the current frame, confirm the pixel values ​​associated with different pixel points in the central picture, and mark them as X i , where i represents different pixels. Based on other pixels around the corresponding pixel, the Sobel algorithm is used to confirm the horizontal and vertical gradients of the pixel, and the horizontal gradient of the pixel is calibrated as H i , the vertical gradient is calibrated as S i ; S22, based on the lateral gradient H confirmed by the corresponding pixel point i And the vertical gradient S i , confirm the comprehensive gradient ZH of this pixel i ,in Based on the comprehensive gradient confirmed by this pixel point, the four groups of pixels associated with the upper and lower positions and the left and right positions of this pixel point are confirmed to generate a set of pixel sequences: Among them SZ i and XZ i Represents the pixel point associated with the upper position and the lower position, whose ZX i and YX i Represents the pixel points associated with the left and right positions, and SZ i , X i and XZ i The three groups of comprehensive gradients are averaged to confirm the first average gradient T1 i , and then ZX i , X i and YX i The three groups of comprehensive gradients are averaged to confirm the second average gradient T2 i ; Based on the confirmed first mean gradient T1 i and the second mean gradient T2 i , lock T1 i and T2 i The lowest common multiple of , and the locked common multiple is used as the gradient feature of this pixel; S23. Based on the different gradient features confirmed by different pixel points, a maximum value is selected from the confirmed groups of gradient features, and the pixel point corresponding to the maximum value is used as a feature point associated with the current frame.

4. The intelligent traffic police mobile video monitoring method according to claim 1, characterized in that: In step 3, the specific sub-steps of confirming the correlation vector are: S31, the next frame of the generated picture is processed in the same manner as in step 2, the central picture is preferentially determined, and then the gradient features associated with each different pixel point in the central picture are confirmed in the same manner as in steps S21-S22; S32, the gradient features associated with the feature points in the previous frame are recorded as upper gradient features, and the gradient features associated with different pixel points in the next frame are recorded as lower gradient features: From the several groups of lower gradient features confirmed in the next frame, identify the pixel points with the same lower gradient features as the upper gradient features. If there are related pixel points, record the existing pixel points as the moving points of the feature points. Take the position of the feature points as the starting point and the position of the moving points as the end point to confirm a group of associated vectors. The confirmed moving points are the feature points of this frame, which are used for confirmation of the associated vectors of the next frame.

5. The intelligent traffic police mobile video monitoring method according to claim 4 is characterized in that: In step S32: If there are no relevant pixel points, the pixel sequence associated with the feature point is confirmed and recorded as the main pixel sequence, and based on the location of the feature point, the same position point is confirmed in the next frame, and the confirmed position point is used as the center of the circle, and then a group of radial circles is confirmed with a radius value of R1, where R1 is a preset value, and the pixel sequences associated with different pixel points in the radial circle are confirmed one by one, and the confirmed pixel sequences are recorded as sub-pixel sequences, and the gradient features at the same sequence position of the main pixel sequence and the sub-pixel sequence are difference processed to confirm the gradient sequence difference at the same sequence position. The difference processing method of each same sequence position is the same, and several groups of gradient sequence difference values ​​belonging to the same sub-pixel sequence are proofread to identify whether the several groups of gradient sequence difference values ​​are consistent. If they are consistent, the pixel point associated with this pixel sequence is marked as a moving point. If they are not consistent, the gradient sequence difference values ​​associated with other sub-pixel sequences are continuously confirmed and proofread until the moving point is confirmed. If the moving point has not been confirmed, the feature point of the current frame is re-confirmed in the same way as step 2; Taking the position of the feature point as the starting point and the position of the moving point as the end point, a set of associated vectors is confirmed. The confirmed moving point is the feature point of the current frame, which is used for confirmation of the associated vector of the next frame.

6. The intelligent traffic police mobile video monitoring method according to claim 1 is characterized in that: In step 4, the specific method of real-time control of the monitoring device is: The monitoring probe is controlled to move in the opposite direction in real time, with the end point of the correlation vector as the starting point and the starting point of the correlation vector as the end point. The monitoring probe is controlled in real time based on the correlation vectors confirmed between different adjacent frames.

7. A smart traffic police mobile video monitoring system, which operates according to the smart traffic police mobile video monitoring method according to claims 1-6, characterized in that: include: The center picture calibration end determines a group of center areas from the total coverage area based on the total coverage area of ​​the pictures monitored by the monitoring equipment, and calibrates the pictures associated with the center areas as the center pictures. The monitoring pictures of different frames are all confirmed to have associated center pictures. The feature point calibration end performs feature point analysis on the central image confirmed in the current frame, confirms the pixel values ​​of different pixels in the central image, and then confirms the gradient features associated with the corresponding pixels. Based on the different gradient features associated with different pixels, feature points are determined from different pixels. The frame vector confirmation end is located at the next group of frame images of the current frame image, and processes the next group of frame images in the same manner as step 2, giving priority to confirming the central image, and confirming the gradient features associated with different pixel points in the central image, locking the feature points confirmed by the previous frame image, and confirming the associated vector based on the position change associated with the feature points; The probe control adjustment end controls the monitoring device in real time based on the correlation vector confirmed between adjacent frame images, so that the monitoring probe inside the monitoring device moves in the opposite direction.

Images