Personnel intrusion visual identification method suitable for hydropower station dam drainage scene

Abstract

The invention discloses a personnel intrusion visual identification method suitable for a hydropower station dam drainage scene. The method comprises the following steps: obtaining a monitoring video image; phase information of the monitoring video image and optical flow information of pixels in the monitoring video image are calculated in sequence; performing equal-area region division on the monitoring video image, and respectively calculating optical flow information of pixels in the region; and respectively calculating Gini impurities in different areas according to the optical flow information of the pixels in the areas to judge the pixel motion confusion degree in the areas and carry out intrusion identification. The personnel intrusion visual identification method suitable for the hydropower station dam drainage scene is not affected by video image data samples in the practical application level, and has the advantages of being low in cost and convenient to popularize and use; in a tiny target identification level, texture features in deep learning are replaced by pixel motion features, and when a tiny target is identified, the method has high identification precision.

Description

Technical Field

[0001] The present invention belongs to the technical field of safety monitoring of hydropower stations, and relates to a human intrusion visual recognition method suitable for a hydropower station dam discharge scenario. Background Art

[0002] As one of the main contents of hydropower station operation and maintenance, hydropower station dam discharge is often sudden due to external factors such as power grid peak regulation, load changes, and water and rainfall conditions. Therefore, it is of great value to timely and accurately identify whether there is human or animal invasion in the downstream river. At present, the downstream river inspection during dam discharge mainly adopts manual inspection, that is, the staff patrols both sides of the river in a specific vehicle. This method is time-consuming and labor-intensive, and the inspection range of the downstream river is large, and the cost of a single inspection is high. In recent years, with the development of computer vision technology, especially with the improvement of image acquisition technology and image processing technology, visual monitoring has gradually become the main inspection method for rivers. However, due to the large monitoring range of the river and the small number of pixels occupied by human targets in the image, the existing monitoring methods based on image target recognition are not effective; on the other hand, the small number of image data samples and insufficient training of deep learning models also lead to low recognition accuracy; in addition, the current deep learning-based model is easily affected by external conditions such as light and illumination, and the recognition results have high uncertainty, resulting in the current monitoring system not only failing to provide convenience for inspection personnel, but also increasing maintenance costs to a certain extent.

[0003] In summary, the existing technology has the problem of low intrusion recognition accuracy due to factors such as the recognition target occupying few pixels and the recognition process being affected by light interference. Summary of the invention

[0004] The purpose of the present invention is to provide a method for visually identifying human intrusion in a hydropower station dam discharge scenario, which solves the problem of low intrusion recognition accuracy in the prior art due to factors such as the small number of pixels occupied by the recognition target and the interference of light during the recognition process.

[0005] The technical solution adopted by the present invention is a method for visually identifying human intrusion in a hydropower station dam discharge scenario, comprising the following steps: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion chaos in the region and perform intrusion identification.

[0006] The present invention is also characterized in that: Step 2 includes: Step 2.1, using Gabor wavelet to convolve the surveillance video image to obtain the surveillance video image phase information; Step 2.2: Calculate the optical flow information of pixels in the surveillance video image based on the phase information of the surveillance video image.

[0007] The convolution process is as follows: , in, Indicates the monitoring video image intensity information. Indicates the monitoring video image phase information, Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the convolution operation, Indicates surveillance video images.

[0008] The Gabor wavelet is shown below: , in, represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the Gabor wavelet function, represents the horizontal coordinate of the Gabor wavelet kernel, represents the vertical coordinate of the Gabor wavelet kernel, represents the wavelength of the Gabor wavelet, represents the direction of the Gabor wavelet, represents the phase shift of Gabor wavelet, represents the standard deviation of the Gabor wavelet Gaussian factor, Represents the spatial aspect ratio of the Gabor wavelet.

[0009] The optical flow information of pixels in the surveillance video image is as follows: , in, Indicates the monitoring video image phase information, Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. Represents the optical flow information in the horizontal direction of the surveillance video image. Represents the optical flow information in the vertical direction of the surveillance video image.

[0010] The calculation formula for the optical flow information of pixels in the area is as follows: , , in, express The optical flow information in the horizontal direction of the surveillance video image at each moment, express The optical flow information in the vertical direction of the surveillance video image at each moment, express Always monitor the phase information of the video image in the 0° direction. express Always monitor the phase information of the video image in the 90° direction. Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinates in the surveillance video image, Indicates the time of monitoring video image.

[0011] Step 4 includes: Step 4.1, counting the distribution probability of the optical flow information of pixels in different regions based on the optical flow information of the pixels in the region; Step 4.2, calculate the Gini impurity in different areas according to the probability of pixel optical flow distribution in different areas; Step 4.3: Determine the degree of pixel motion disorder in the region based on the Gini impurity in the region and perform intrusion identification; When the Gini impurity in the region is 0, the optical flow values ​​of all pixels in the region are consistent, that is, there is no intrusion; when the Gini impurity in the region is between 0-1, the optical flow values ​​of all pixels in the region are inconsistent, that is, there is intrusion.

[0012] The calculation formula for the distribution probability of pixel optical flow information in the area is as follows: , in, Indicates area The probability of the distribution of the inner pixel optical flow information, Indicates a certain area in the surveillance video image. N Indicates area The number of pixels within Indicates area Horizontal distribution of internal pixels 、 Indicates area The vertical distribution of the inner pixels.

[0013] The calculation formula of Gini impurity in a region is as follows: , in, Indicates area The Gini impurity of Indicates area Intra-pixel optical flow information distribution probability.

[0014] The formula for determining the degree of pixel motion disorder in the region is as follows: , in, Indicates Regions The Gini impurity of Indicates the first Regions, Indicates the total number of areas after the surveillance video image is divided.

[0015] The beneficial effects of the present invention are: at the practical application level, the present invention is not affected by video image data samples, and has the advantages of low cost and easy promotion and use; at the small target recognition level, the present invention replaces the texture features in deep learning with pixel motion features, and has higher recognition accuracy when identifying small targets; in terms of recognition and anti-interference, the present invention uses image phase features instead of image grayscale features, and has higher recognition accuracy under conditions of weak light and poor lighting; in terms of convenience, compared with the current recognition method based on deep learning model, the present invention does not need to use data to train the model, does not need to label video sample data, and has relatively high deployment efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a flow chart of the method for visually identifying human intrusion in a hydropower station dam discharge scenario according to the present invention. DETAILED DESCRIPTION

[0017] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0018] Applicable to the human intrusion visual recognition method in the hydropower station dam discharge scene, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 2.1, using Gabor wavelet to convolve the surveillance video image to obtain the surveillance video image phase information; The convolution process is as follows: , in, Indicates the monitoring video image intensity information. Indicates the monitoring video image phase information, Indicates the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the convolution operation, Indicates a surveillance video image; The Gabor wavelet is shown below: , in, represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the Gabor wavelet function, represents the horizontal coordinate of the Gabor wavelet kernel, represents the vertical coordinate of the Gabor wavelet kernel, represents the wavelength of the Gabor wavelet, represents the direction of the Gabor wavelet, represents the phase shift of Gabor wavelet, represents the standard deviation of the Gabor wavelet Gaussian factor, represents the spatial aspect ratio of the Gabor wavelet; Step 2.2, calculating the optical flow information of pixels in the surveillance video image according to the phase information of the surveillance video image; The optical flow information of pixels in the surveillance video image is as follows: , in, Indicates the monitoring video image phase information, Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. Represents the optical flow information in the horizontal direction of the surveillance video image. Represents the optical flow information in the vertical direction of the surveillance video image; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; The calculation formula for the optical flow information of pixels in the area is as follows: , , in, express The optical flow information in the horizontal direction of the surveillance video image at each moment, express The optical flow information in the vertical direction of the surveillance video image at each moment, express Always monitor the phase information of the video image in the 0° direction. express Always monitor the phase information of the video image in the 90° direction. Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinates in the surveillance video image, Indicates the time when the video image is monitored; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion disorder in the region and perform intrusion identification; Step 4.1, counting the distribution probability of the optical flow information of pixels in different regions based on the optical flow information of the pixels in the region; The calculation formula for the distribution probability of pixel optical flow information in the area is as follows: , in, Indicates area The probability of the distribution of the inner pixel optical flow information, Indicates a certain area in the surveillance video image. N Indicates area The number of pixels within Indicates area Horizontal distribution of internal pixels 、 Indicates area The distribution of internal pixels in the vertical direction; Step 4.2, calculate the Gini impurity in different areas according to the probability of pixel optical flow distribution in different areas; The calculation formula of Gini impurity in a region is as follows: , in, Indicates area The Gini impurity of Indicates area Intra-pixel optical flow information distribution probability; Step 4.3: Determine the degree of pixel motion disorder in the region based on the Gini impurity in the region and perform intrusion identification; When the Gini impurity in the region is 0, the optical flow values ​​of all pixels in the region are consistent, that is, there is no intrusion; when the Gini impurity in the region is between 0-1, the optical flow values ​​of all pixels in the region are inconsistent, that is, there is intrusion; The formula for determining the degree of pixel motion disorder in the region is as follows: , in, Indicates Regions The Gini impurity of Indicates the first Regions, Indicates the total number of areas after the surveillance video image is divided.

[0019] In computer vision theory, a grayscale image is composed of image phase and image intensity, wherein the image phase describes the operation information of the structure within the image, and the image intensity represents the brightness information of the image. The present invention uses image phase information for intrusion recognition, which is more robust than the structural motion measurement method based on grayscale information.

[0020] Example 1 This embodiment proposes a human intrusion visual recognition method suitable for the hydropower station dam discharge scene, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion chaos in the region and perform intrusion identification.

[0021] Example 2 This embodiment proposes a human intrusion visual recognition method suitable for the hydropower station dam discharge scene, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 2.1, using Gabor wavelet to convolve the surveillance video image to obtain the surveillance video image phase information; Step 2.2, calculating the optical flow information of pixels in the surveillance video image according to the phase information of the surveillance video image; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion chaos in the region and perform intrusion identification.

[0022] Example 3 This embodiment proposes a human intrusion visual recognition method suitable for the hydropower station dam discharge scene, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 2.1, using Gabor wavelet to convolve the surveillance video image to obtain the surveillance video image phase information; The convolution process is as follows: , in, Indicates the monitoring video image intensity information. Indicates the monitoring video image phase information, Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the convolution operation, Indicates a surveillance video image; The Gabor wavelet is shown below: , in, represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the Gabor wavelet function, represents the horizontal coordinate of the Gabor wavelet kernel, represents the vertical coordinate of the Gabor wavelet kernel, represents the wavelength of the Gabor wavelet, represents the direction of the Gabor wavelet, represents the phase shift of Gabor wavelet, represents the standard deviation of the Gabor wavelet Gaussian factor, represents the spatial aspect ratio of the Gabor wavelet; Step 2.2, calculating the optical flow information of pixels in the surveillance video image according to the phase information of the surveillance video image; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion chaos in the region and perform intrusion identification.

[0023] Example 4 This embodiment proposes a human intrusion visual recognition method suitable for the hydropower station dam discharge scene, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 2.1, using Gabor wavelet to convolve the surveillance video image to obtain the surveillance video image phase information; Step 2.2, calculating the optical flow information of pixels in the surveillance video image according to the phase information of the surveillance video image; The optical flow information of pixels in the surveillance video image is as follows: , in, Indicates the monitoring video image phase information, Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. Represents the optical flow information in the horizontal direction of the surveillance video image. Represents the optical flow information in the vertical direction of the surveillance video image; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion chaos in the region and perform intrusion identification.

[0024] Example 5 This embodiment proposes a human intrusion visual recognition method suitable for a hydropower station dam discharge scenario, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the monitoring video image phase information and the optical flow information of the pixels in the monitoring video image in sequence; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; The calculation formula for the optical flow information of pixels in the area is as follows: , , in, express The optical flow information in the horizontal direction of the surveillance video image at each moment, express The optical flow information in the vertical direction of the surveillance video image at each moment, express Always monitor the phase information of the video image in the 0° direction. express Always monitor the phase information of the video image in the 90° direction. Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinates in the surveillance video image, Indicates the time of monitoring video image; Step 4: Calculate the Gini impurity of different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion chaos in the region and perform intrusion identification.

[0025] Example 6 This embodiment proposes a human intrusion visual recognition method suitable for a hydropower station dam discharge scenario, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the monitoring video image phase information and the optical flow information of the pixels in the monitoring video image in sequence; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion disorder in the region and perform intrusion identification; Step 4.1, counting the distribution probability of the optical flow information of pixels in different regions based on the optical flow information of the pixels in the region; Step 4.2, calculate the Gini impurity in different areas according to the probability of pixel optical flow distribution in different areas; Step 4.3: Determine the degree of pixel motion disorder in the region based on the Gini impurity in the region and perform intrusion identification; When the Gini impurity in the region is 0, the optical flow values ​​of all pixels in the region are consistent, that is, there is no intrusion; when the Gini impurity in the region is between 0-1, the optical flow values ​​of all pixels in the region are inconsistent, that is, there is intrusion.

[0026] Example 7 This embodiment proposes a human intrusion visual recognition method suitable for the hydropower station dam discharge scene, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion disorder in the region and perform intrusion identification; Step 4.1, counting the distribution probability of the optical flow information of pixels in different regions based on the optical flow information of the pixels in the region; The calculation formula for the distribution probability of pixel optical flow information in the area is as follows: , in, Indicates area The probability of the distribution of the inner pixel optical flow information, Indicates a certain area in the surveillance video image. N Indicates area The number of pixels within Indicates area Horizontal distribution of internal pixels 、 Indicates area The distribution of internal pixels in the vertical direction; Step 4.2, calculate the Gini impurity in different areas according to the probability of pixel optical flow distribution in different areas; Step 4.3: Determine the degree of pixel motion disorder in the region based on the Gini impurity in the region and perform intrusion identification; When the Gini impurity in the region is 0, the optical flow values ​​of all pixels in the region are consistent, that is, there is no intrusion; when the Gini impurity in the region is between 0-1, the optical flow values ​​of all pixels in the region are inconsistent, that is, there is intrusion.

[0027] Example 8 This embodiment proposes a human intrusion visual recognition method suitable for a hydropower station dam discharge scenario, such as Figure 1 As shown, the following steps are included: Step 1: Obtain surveillance video images; Step 2: Calculate the monitoring video image phase information and the optical flow information of the pixels in the monitoring video image in sequence; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion disorder in the region and perform intrusion identification; Step 4.1, counting the distribution probability of the optical flow information of pixels in different regions based on the optical flow information of the pixels in the region; Step 4.2, calculate the Gini impurity in different areas according to the probability of pixel optical flow distribution in different areas; The calculation formula of Gini impurity in a region is as follows: , in, Indicates area The Gini impurity of Indicates area Intra-pixel optical flow information distribution probability; Step 4.3: Determine the degree of pixel motion disorder in the region based on the Gini impurity in the region and perform intrusion identification; When the Gini impurity in the region is 0, the optical flow values ​​of all pixels in the region are consistent, that is, there is no intrusion; when the Gini impurity in the region is between 0-1, the optical flow values ​​of all pixels in the region are inconsistent, that is, there is intrusion; The formula for determining the degree of pixel motion disorder in the region is as follows: , in, Indicates Regions The Gini impurity of Indicates the first Regions, Indicates the total number of areas after the surveillance video image is divided.

[0028] The present invention breaks through the problems of high training cost of current deep learning-based target recognition algorithms, low precision in identifying small targets in large scenes, and interference of light in identification results, and provides technical support for foreign body intrusion in the process of water discharge in hydropower station dams. The present invention considers the dynamic characteristics of human and animal intrusion processes, relies on river video monitoring points, and proposes a human intrusion visual recognition method suitable for hydropower station dam water discharge scenes. Specifically, by acquiring video images of river video monitoring points, an optical flow calculation method based on image phase is proposed to calculate the optical flow information of each pixel in the image, and further partition the monitoring area. Combined with the optical flow information, the Gini impurity of pixel motion in different areas is calculated, the activity level of the local area is judged, and the degree of chaos is used to identify whether the local area has formed an invasion. The present invention can effectively improve the recognition accuracy of human and animal intrusion.

Claims

1. A visual recognition method for human intrusion in a hydropower station dam discharge scenario, characterized in that: The following steps are involved: Step 1: Obtain surveillance video images; Step 2: Calculate the phase information of the surveillance video image and the optical flow information of the pixels in the surveillance video image in sequence; Step 3: Divide the surveillance video image into regions of equal area and calculate the optical flow information of the pixels in each region; Step 4: Calculate the Gini impurity in different regions based on the optical flow information of the pixels in the region to determine the degree of pixel motion chaos in the region and perform intrusion identification.

2. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 1 is characterized in that: The second step comprises: Step 2.1, using Gabor wavelet to convolve the surveillance video image to obtain the surveillance video image phase information; Step 2.2: Calculate the optical flow information of pixels in the surveillance video image based on the phase information of the surveillance video image.

3. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 2 is characterized in that: The convolution process is as follows: in, Indicates the monitoring video image intensity information. Indicates the monitoring video image phase information, Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the convolution operation, Indicates surveillance video images.

4. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 3 is characterized in that: The Gabor wavelet is shown as follows: ， in, represents Gabor wavelet, represents the real part of the Gabor wavelet, represents the imaginary part of the Gabor wavelet, represents the Gabor wavelet function, represents the horizontal coordinate of the Gabor wavelet kernel, represents the vertical coordinate of the Gabor wavelet kernel, represents the wavelength of the Gabor wavelet, represents the direction of the Gabor wavelet, represents the phase shift of Gabor wavelet, represents the standard deviation of the Gabor wavelet Gaussian factor, Represents the spatial aspect ratio of the Gabor wavelet.

5. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 2 is characterized in that: The optical flow information of the pixels in the surveillance video image is as follows: ， in, Indicates the monitoring video image phase information, Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinate of the surveillance video image, Indicates the time when the video image is monitored. Represents the optical flow information in the horizontal direction of the surveillance video image. Represents the optical flow information in the vertical direction of the surveillance video image.

6. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 1 is characterized in that: The calculation formula for the optical flow information of pixels in the region is as follows: ， ， in, express The optical flow information in the horizontal direction of the surveillance video image at each moment, express The optical flow information in the vertical direction of the surveillance video image at each moment, express Always monitor the phase information of the video image in the 0° direction. express Always monitor the phase information of the video image in the 90° direction. Represents the horizontal pixel coordinates of the surveillance video image. Represents the vertical pixel coordinates in the surveillance video image, Indicates the time of monitoring video image.

7. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 1 is characterized in that: The fourth step comprises: Step 4.1, counting the distribution probability of the optical flow information of pixels in different regions based on the optical flow information of the pixels in the region; Step 4.2, calculate the Gini impurity in different areas according to the probability of pixel optical flow distribution in different areas; Step 4.3: Determine the degree of pixel motion disorder in the region based on the Gini impurity in the region and perform intrusion identification; When the Gini impurity in the region is 0, the optical flow values ​​of all pixels in the region are consistent, that is, there is no intrusion; when the Gini impurity in the region is between 0-1, the optical flow values ​​of all pixels in the region are inconsistent, that is, there is intrusion.

8. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 7 is characterized in that: The calculation formula for the distribution probability of pixel optical flow information in the region is as follows: ， in, Indicates area The probability of the distribution of the inner pixel optical flow information, Indicates a certain area in the surveillance video image. N Indicates area The number of pixels within Indicates area Horizontal distribution of internal pixels 、 Indicates area The vertical distribution of the inner pixels.

9. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 7, characterized in that: The calculation formula of Gini impurity in the region is as follows: ， in, Indicates area The Gini impurity of Indicates area Intra-pixel optical flow information distribution probability.

10. The method for visually identifying human intrusion in a hydropower station dam discharge scenario according to claim 7, characterized in that: The formula for determining the degree of pixel motion disorder in the region is as follows: ， in, Indicates Regions The Gini impurity of Indicates the first Regions, Indicates the total number of areas after the surveillance video image is divided.

Images