Method for assisting manual light following by utilizing machine vision and light control system

Abstract

The invention discloses a method for assisting manual light following by utilizing machine vision and a light control system, and the method comprises the steps: shooting an image containing a target human body by utilizing a camera with a known pose, calculating the prediction state information containing a prediction coordinate position of the target human body at the tth moment according to the motion state information of the target human body at the (t-1) th moment, and taking the target human body position manually input by the user at the t-th moment as a real coordinate position, combining the real coordinate position with the predicted state information to obtain the motion state information of the target human body at the t-th moment, and calculating the space coordinate position of the target human body according to the estimated coordinate position contained in the motion state information in combination with the camera parameters. And sending to the follow spot lamp for tracking. And the background corrects the input real coordinate position of the target human body at the current moment according to the predicted coordinate position, and automatically performs fine adjustment on the manual light tracking result of the user, so that the manual light tracking positioning of the user is more accurate, the light spot is not easy to shake, the continuity is better, and the user is not easy to lose.

Description

Technical Field

[0001] The present invention relates to the technical field of chasing light, and more specifically, to a method for manually chasing light using machine vision assistance and a lighting control system. Background Art

[0002] In the current performing arts lighting system, the spotlight follows the target person on the stage. Generally, the lighting technician needs to manually control the spotlight to rotate so that the light spot follows the target person and illuminates him. Since the spotlight is generally far away from the stage, the manual tracking accuracy is not easy to control. Another common method is that the target person wears a special active tag (such as a UWB tag) or a passive tag (such as an infrared marker), but on the one hand, the target person may not be convenient to wear the tag in certain occasions, and on the other hand, the data transmitted by the active tag may be interfered by the human body or the surrounding scene, and the passive tag may be blocked, resulting in inaccurate positioning of the target person or jitter in positioning. Another method is to use a camera to identify the target person and track it, but this method is easy to lose the target person when multiple people move in an interlaced manner, multiple people have similar figures, or the light is dim. For this reason, a label-free human body high-precision continuous and stable light tracking method is needed. Summary of the invention

[0003] In order to overcome at least one defect of the above-mentioned prior art, the present invention provides a method of using machine vision to assist manual light chasing, which can effectively solve the problem of manual light chasing being prone to jitter and having poor accuracy.

[0004] To solve the above technical problems, the technical solution adopted by the present invention is: a method of using machine vision to assist manual light chasing, comprising using a camera with a known posture to shoot an image containing a target human body, calculating the predicted state information of the target human body at the t moment including the predicted coordinate position based on the motion state information of the target human body at the t-1 moment, taking the target human body position at the t moment manually input by the user as the real coordinate position, combining the real coordinate position with the predicted state information to obtain the motion state information of the target human body at the t moment, and calculating the spatial coordinate position of the target human body on the stage based on the estimated coordinate position contained in the motion state information of the target human body at the t moment, combined with the posture and internal parameters of the camera, and sending it to the chase light for tracking.

[0005] The method of using machine vision to assist manual light chasing predicts the predicted coordinate position at the current moment based on the motion state information of the target body at the previous moment, and combines it with the target body position at the current moment input by the user as the real coordinate position to obtain the estimated coordinate position of the target body at the current moment. The chasing light performs light chasing based on the estimated coordinate position. In other words, the background will correct the input real coordinate position of the target body at the current moment according to the predicted coordinate position, and automatically fine-tune the result of the user's manual light chasing, so that the user's manual light chasing positioning is more accurate, and the light spot will not shake randomly due to the slight movement of the input device, and the manual light chasing has better continuity and is not easy to lose the target body.

[0006] Furthermore, the Kalman filter is used to combine the real coordinate position with the predicted state information to obtain the motion state information of the target human body at the tth moment. As a mature data processing algorithm, the Kalman filter can gradually correct and predict the data, so that the results are closer to the actual situation and more accurate.

[0007] Furthermore, a human body rectangular frame is identified from an image of a target human body captured by a camera, and the predicted coordinate position of the target human body includes the predicted center point position of the human body rectangular frame, and the target human body position manually input by the user is used as the center point position of the real human body rectangular frame. That is, the center point position of the human body rectangular frame is input and calibrated, and the center point position of the target human body has a small movement range, which is easy to select and more in line with the user's operating habits.

[0008] Furthermore, the predicted coordinate position of the target human body also includes the predicted coordinates of the lower midpoint of the human body rectangular frame, and the real coordinate position of the target human body also includes the real coordinates of the lower midpoint of the human body rectangular frame detected at the t-th time. When the real coordinate position is combined with the predicted state information, the coordinates of the lower midpoint of the human body rectangular frame and the center point position of the human body rectangular frame are considered at the same time, so as to obtain the motion state information of the target human body at the t-th time. The coordinates of the lower midpoint of the human body rectangular frame and the center point position of the human body rectangular frame are considered at the same time, and prediction and correction are performed together, and two observation quantities are introduced to constrain each other, and the motion state information of the target human body is jointly derived, and the result is more accurate.

[0009] Furthermore, the motion state information of the target human body at time t-1 is X t-1 =(x,y,δ x ,δ y ,v x ,v y ) T , where (x, y) is the coordinate of the middle point of the lower side of the human body rectangular frame at time t-1, δ x ,δ y is the offset from the bottom midpoint of the human body rectangle to the center of the human body rectangle at time t-1, vx ,v y represents the velocity of the middle point of the lower edge of the human body rectangle at time t-1, where at the first moment, v x ,v y are all 0. Generally, after the human body rectangle is identified, the coordinates of the upper left corner and the lower right corner of the human body rectangle are obtained, and (x, y) and δ can be calculated based on this. x ,δ y , and use (x,y) and δ x ,δ y The center point position of the human body rectangular frame can be calculated to facilitate the subsequent calculation of the predicted state information of the target human body at the tth moment.

[0010] Furthermore, the predicted state information of the target human body at the tth moment is: (1), A is the transfer matrix of the lighting control system, and , Δt is the time interval between the tth moment and the t-1th moment. The predicted coordinate position of the target human body at the tth moment can be obtained according to the predicted state information of the target human body at the tth moment. By default, only the coordinate of the middle point of the lower edge of the human body rectangular frame has changed compared to the t-1th moment, and the offset δ from the middle point of the lower edge of the human body rectangular frame to the center point of the human body rectangular frame x ,δ y Keep constant, reduce variables, and facilitate calculation.

[0011] Furthermore, the real coordinate position of the target human body at the t moment manually input by the user is combined with the predicted state information to obtain the motion state information of the target human body at the t moment, specifically, Calculate the prediction error covariance matrix between the predicted coordinate position and the true coordinate position at time t: (2), Among them, Q is a 6*6 diagonal matrix, P t-1 is the error covariance matrix between the estimated coordinate position and the true coordinate position at time t-1; Then find the Kalman gain: (3), Where H is the observation matrix, and , R is a 4*4 diagonal matrix; Finally, calculate the motion state information of the target human body at time t: (4), Among them, Zt is the real coordinate position of the target body at the tth moment, which is equal to , is the coordinate of the lower midpoint of the real human body rectangular frame detected at the tth moment, The target human body position manually input by the user at time t; Update the error covariance matrix between the estimated coordinate position and the true coordinate position at time t: (5) Where I is the identity matrix.

[0012] After solving formulas (1), (2), (3), and (4) in sequence, we can obtain the motion state information X of the target human body at the tth moment by combining the real coordinate position and the predicted state information after considering the coordinates of the midpoint of the lower side of the human body rectangular frame and the center point position of the human body rectangular frame at the same time: t , which includes the estimated coordinate position, and the error covariance matrix P between the estimated coordinate position and the true coordinate position at time t is updated according to formula (5) t , is to calculate the prediction error covariance matrix `P between the predicted coordinate position and the true coordinate position at the t+1th moment t+1 Be prepared.

[0013] Furthermore, the estimated coordinates of the lower midpoint of the human body rectangular frame in the estimated coordinate position are combined with the camera's position and internal parameters to calculate the spatial coordinate position of the target human body on the stage, and sent to the follow-up light for tracking. When the follow-up light shines on the target human body, it generally shines on the ground where the target human body is located. Therefore, letting the follow-up light track according to the estimated coordinates of the lower midpoint of the human body rectangular frame will achieve a better stage effect.

[0014] Furthermore, the image of the target human body captured by the camera is displayed on the display screen, and the user manually inputs the position of the target human body at the tth moment as the real coordinate position on the display screen through the cursor. Using the display screen, the user can intuitively and quickly input the position of the target human body at the tth moment, and can predict the position of the target human body at the next moment to a certain extent, thereby improving the input speed.

[0015] Furthermore, the cursor position is controlled by the user through a mouse, a joystick or a touch screen, and the cursor position at the tth moment is used as the target human body position. The mouse, the joystick or the touch screen is quick and convenient to use and conforms to the user's operating habits.

[0016] The present invention also provides a lighting control system, which runs any of the aforementioned methods of using machine vision to assist manual light chasing, including a camera with a known posture for shooting an image containing a target human body, an input device for a user to manually input the position of the target human body at the tth moment, and a server and a light chasing light. The server calculates the predicted state information of the target human body at the tth moment including the predicted coordinate position based on the motion state information of the target human body at the t-1th moment, and combines the real coordinate position with the predicted state information to obtain the motion state information of the target human body at the tth moment, and calculates the spatial coordinate position of the target human body on the stage by combining the estimated coordinate position with the posture and internal parameters of the camera, and the light chasing light tracks the target human body according to the spatial coordinate position. The lighting control system uses machine vision to automatically assist manual light chasing, so that when users chase the light manually, the light chasing is more accurate and smooth, which invisibly improves the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the method of using machine vision to assist manual light chasing according to the present invention.

[0018] Figure 2 It is a schematic diagram of an image displayed on the display screen of the present invention.

[0019] Figure 3 It is a structural schematic diagram of the lighting control system of the present invention. DETAILED DESCRIPTION

[0020] The drawings are only for illustrative purposes and cannot be construed as limiting the present invention. To better illustrate the present embodiment, some parts of the drawings may be omitted, enlarged, or reduced, and do not represent the size of the actual product. For those skilled in the art, it is understandable that some well-known structures and their descriptions may be omitted in the drawings. The positional relationships described in the drawings are only for illustrative purposes and cannot be construed as limiting the present invention.

[0021] like Figures 1 to 3 The present invention provides a method for manual light chasing using machine vision, comprising: using a camera with known posture (position and attitude) to capture an image containing a target human body; calculating predicted state information containing predicted coordinate positions of the target human body at the t-1th moment (t≥2) according to the motion state information of the target human body; using the position of the target human body at the t-1th moment manually input by a user as the real coordinate position; combining the real coordinate position with the predicted state information to obtain the motion state information of the target human body at the tth moment; and calculating the spatial coordinate position of the target human body on the stage according to the estimated coordinate position contained in the motion state information of the target human body at the tth moment in combination with the posture and internal parameters of the camera, and sending the calculated spatial coordinate position to a chasing light for tracking.

[0022] The method of using machine vision to assist manual light chasing predicts the predicted coordinate position at the current moment based on the motion state information of the target body at the previous moment, and combines it with the target body position at the current moment input by the user as the real coordinate position to obtain the estimated coordinate position of the target body at the current moment. The chasing light performs light chasing based on the estimated coordinate position. In other words, the background will correct the input real coordinate position of the target body at the current moment according to the predicted coordinate position, and automatically fine-tune the result of the user's manual light chasing, so that the user's manual light chasing positioning is more accurate, and the light spot will not shake randomly due to the slight movement of the input device, and the manual light chasing has better continuity and is not easy to lose the target body.

[0023] It should be noted that in this application, the motion state information and the predicted state information generally include position and speed information, and in other embodiments, may also include acceleration information. The real coordinate position at the t moment may include other real position information in addition to the target human body position manually input by the user.

[0024] In this embodiment, the camera's posture is calibrated by Zhang's calibration method, and then the position information of the target human body is calculated in the image coordinate system, and finally the motion state information of the target human body in the camera coordinate system at the tth moment is calculated, and then the spatial coordinate position of the target human body on the stage (that is, the position in reality, the coordinate system generally takes the center of the stage as the zero point, the ground of the stage as the XY plane, and the Z-axis direction perpendicular to the stage upward) can be mapped by the epipolar geometry constraint algorithm, so that the follow spot can track the target human body.

[0025] In a preferred embodiment of the present invention, a Kalman filter is used to combine the real coordinate position with the predicted state information to obtain the motion state information of the target human body at time t. As a mature data processing algorithm, Kalman filtering can gradually correct and predict the data, so that the results are closer to the actual situation and more accurate.

[0026] In a preferred embodiment of the present invention, a human body rectangular frame is identified from an image of a target human body captured by a camera, the predicted coordinate position of the target human body includes the predicted center point position of the human body rectangular frame, and the target human body position manually input by the user is used as the center point position of the real human body rectangular frame. That is, the center point position of the human body rectangular frame is input and calibrated, and the center point position of the target human body has a small movement range, which is easy to select and more in line with the user's operating habits.

[0027] In this embodiment, the v7 version of the YOLO algorithm is used to identify the human body rectangular frame from the image containing the target human body taken by the camera. In other embodiments, other algorithms or versions may also be used.

[0028] In a preferred embodiment of the present invention, the predicted coordinate position of the target human body also includes the predicted coordinates of the lower midpoint of the human body rectangular frame, and the real coordinate position of the target human body also includes the real coordinates of the lower midpoint of the human body rectangular frame detected at the t-th time. When the real coordinate position is combined with the predicted state information, the coordinates of the lower midpoint of the human body rectangular frame and the center point position of the human body rectangular frame are considered at the same time, so as to obtain the motion state information of the target human body at the t-th time. The coordinates of the lower midpoint of the human body rectangular frame and the center point position of the human body rectangular frame are considered at the same time, and prediction and correction are performed together, and two observation quantities are introduced, which are mutually constrained, and the motion state information of the target human body is jointly derived, and the result is more accurate.

[0029] In a preferred embodiment of the present invention, the motion state information of the target human body at time t-1 is X t-1 =(x,y,δ x ,δ y ,v x ,v y ) T , where (x, y) is the coordinate of the middle point of the lower side of the human body rectangular frame at time t-1, δ x ,δ y is the offset from the bottom midpoint of the human body rectangle to the center of the human body rectangle at time t-1, v x ,v y represents the velocity of the middle point of the lower edge of the human body rectangle at time t-1, where at the first moment, v x ,v y are all 0. Generally, after the human body rectangle is identified, the coordinates of the upper left corner and the lower right corner of the human body rectangle are obtained, and (x, y) and δ can be calculated based on this. x ,δ y , and use (x,y) and δ x ,δ y The center point position of the human body rectangular frame can be calculated to facilitate the subsequent calculation of the predicted state information of the target human body at the tth moment.

[0030] If the human body rectangle obtained at the first moment is (x l ,y t , x r ,y b ) when [where (x l ,y t ) is the coordinate of the upper left corner of the human body rectangle, (x r ,y b ) is the coordinate of the lower right corner of the human body rectangular box], then the motion state information of the target human body at the first moment is X1=(x,y,δ x ,δ y ,v x ,v y ) T=[x l +(x r -x l ) / 2,y b ,0,(y t -y b ) / 2,0,0] T .

[0031] In a preferred embodiment of the present invention, the predicted state information of the target human body at time t is: (1), A is the transfer matrix of the lighting control system, and , Δt is the time interval between the tth moment and the t-1th moment. The predicted coordinate position of the target human body at the tth moment can be obtained according to the predicted state information of the target human body at the tth moment. By default, only the coordinate of the middle point of the lower edge of the human body rectangular frame has changed compared to the t-1th moment, and the offset δ from the middle point of the lower edge of the human body rectangular frame to the center point of the human body rectangular frame x ,δ y Keep constant, reduce variables, and facilitate calculation.

[0032] In a preferred embodiment of the present invention, the real coordinate position of the target human body at the t moment manually input by the user is combined with the predicted state information to obtain the motion state information of the target human body at the t moment, specifically, Calculate the prediction error covariance matrix between the predicted coordinate position and the true coordinate position at time t: (2), Among them, Q is a 6*6 diagonal matrix, P t-1 is the error covariance matrix between the estimated coordinate position and the true coordinate position at time t-1; Then find the Kalman gain: (3), Where H is the observation matrix, and , R is a 4*4 diagonal matrix; Finally, calculate the motion state information of the target human body at time t: (4), Among them, Zt is the real coordinate position of the target body at the tth moment, which is equal to , is the coordinate of the lower midpoint of the real human body rectangular frame detected at the tth moment, The target human body position manually input by the user at time t; Update the error covariance matrix between the estimated coordinate position and the true coordinate position at time t: (5) Where I is the identity matrix.

[0033] After solving formulas (1), (2), (3), and (4) in sequence, we can obtain the motion state information X of the target human body at the tth moment by combining the real coordinate position and the predicted state information after considering the coordinates of the midpoint of the lower side of the human body rectangular frame and the center point position of the human body rectangular frame at the same time: t , which includes the estimated coordinate position, and the error covariance matrix P between the estimated coordinate position and the true coordinate position at time t is updated according to formula (5) t , is to calculate the prediction error covariance matrix `P between the predicted coordinate position and the true coordinate position at the t+1th moment t+1 Be prepared.

[0034] The values ​​of the diagonal matrices Q and R are calculated based on the height of the human body rectangular frame at the corresponding moment.

[0035] It should be noted that, in this embodiment, it is considered that the error covariance matrix P0 between the estimated coordinate position and the true coordinate position at the first moment is a 6*6 unit matrix.

[0036] In a preferred embodiment of the present invention, the estimated coordinate position of the middle point of the lower side of the human body rectangular frame is combined with the camera's position and internal parameters to calculate the spatial coordinate position of the target human body on the stage, and sent to the follow-up light for tracking. When the follow-up light is directed to the target human body, it generally illuminates the ground where the target human body is located. Therefore, the follow-up light is tracked according to the estimated coordinate position of the middle point of the lower side of the human body rectangular frame, and the stage effect is better.

[0037] In a preferred embodiment of the present invention, the image of the target human body captured by the camera is displayed on the display screen, and the user manually inputs the position of the target human body at the tth moment as the real coordinate position on the display screen through the cursor. Using the display screen, the user can intuitively and quickly input the position of the target human body at the tth moment, and can predict the position of the target human body at the next moment to a certain extent, thereby improving the input speed.

[0038] Preferably, the number of cameras is 3 or more, and the display screen displays the picture taken by at least one of the cameras for the user to manually input the target human body position at the tth moment on the display screen through the cursor.

[0039] In a preferred embodiment of the present invention, it is characterized in that the cursor position is controlled by the user through a mouse, a rocker or a touch screen, and the cursor position at the tth moment is used as the target human body position. The mouse, the rocker or the touch screen is quick and convenient to use, and conforms to the user's operating habits. In this embodiment, the user uses a mouse, a rocker or a touch screen to control the cursor to follow the movement of the target human body, and the cursor position at the tth moment (the cursor may not necessarily stay at this position, but may only be at this position at this moment) is used as the target human body position. In this embodiment, a rocker is used to control the movement of the cursor.

[0040] like Figure 3 The present invention also provides a lighting control system, which runs any of the aforementioned methods of using machine vision to assist manual light chasing, including a camera with a known posture for shooting an image containing a target human body, an input device for a user to manually input the position of the target human body at the tth moment, and a server and a chasing light. The server calculates the predicted state information of the target human body at the tth moment including the predicted coordinate position based on the motion state information of the target human body at the t-1th moment, and combines the real coordinate position with the predicted state information to obtain the motion state information of the target human body at the tth moment, and calculates the spatial coordinate position of the target human body on the stage by combining the estimated coordinate position with the posture and internal parameters of the camera, and the chasing light tracks the target human body according to the spatial coordinate position. The lighting control system uses machine vision to automatically assist manual light chasing, so that when users chase the light manually, the chasing light is more accurate and smooth, which invisibly improves the user experience.

[0041] In this embodiment, the server converts the spatial coordinate position of the target human body on the stage into an angle control command of the follow spot, and sends it to the follow spot via protocols such as DMX512 or Artnet.

[0042] The lighting control system further includes a display screen. The user uses an input device to directly input the body position by controlling the cursor on the display screen. Compared with inputting by looking at the real stage, the user does not need to convert the direction to be consistent with the light, which is more convenient to input and can accurately consider the relative position with the follow-spot light.

[0043] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for manually tracking light using machine vision, characterized in that: The method includes using a camera with a known posture to capture an image containing a target human body, calculating predicted state information containing predicted coordinate positions of the target human body at moment t based on the motion state information of the target human body at moment t-1, taking the target human body position at moment t manually input by a user as the real coordinate position, combining the real coordinate position with the predicted state information to obtain the motion state information of the target human body at moment t, calculating the spatial coordinate position of the target human body on the stage based on the estimated coordinate position contained in the motion state information of the target human body at moment t, combining the posture of the camera and internal parameters, and sending it to a follow spot for tracking.

2. The method of using machine vision to assist manual light tracking according to claim 1, characterized in that: The Kalman filter method is used to combine the real coordinate position with the predicted state information to obtain the motion state information of the target human body at the tth moment.

3. The method for manually chasing light using machine vision as claimed in claim 1 or 2, characterized in that: A human body rectangular frame is identified from an image of a target human body captured by a camera. The predicted coordinate position of the target human body includes the predicted center point position of the human body rectangular frame, and the target human body position manually input by the user is used as the center point position of the real human body rectangular frame.

4. The method of using machine vision to assist manual light tracking according to claim 3, characterized in that: The predicted coordinate position of the target human body also includes the predicted lower midpoint coordinates of the human body rectangular frame, and the real coordinate position of the target human body also includes the real lower midpoint coordinates of the human body rectangular frame detected at the tth moment. When the real coordinate position is combined with the predicted state information, the lower midpoint coordinates of the human body rectangular frame and the center point position of the human body rectangular frame are considered at the same time, thereby obtaining the motion state information of the target human body at the tth moment.

5. The method of using machine vision to assist manual light tracking according to claim 4, characterized in that: The motion state information of the target human body at the t-1th moment is X t-1 =(x,y,δ x ,δ y ,v x ,v y ) T , where (x, y) is the coordinate of the middle point of the lower side of the human body rectangular frame at time t-1, δ x ,δ y is the offset from the bottom midpoint of the human body rectangle to the center of the human body rectangle at time t-1, v x ,v y represents the velocity of the middle point of the lower edge of the human body rectangle at time t-1, where at the first moment, v x ,v y Both are 0.

6. The method of using machine vision to assist manual light tracking according to claim 5, characterized in that: The predicted state information of the target human body at the tth moment is: （1）， A is the transfer matrix of the lighting control system, and , Δt is the time interval between the tth moment and the t-1th moment. According to the predicted state information of the target human body at the tth moment, the predicted coordinate position of the target human body at the tth moment can be obtained.

7. The method of using machine vision to assist manual light tracking according to claim 6, characterized in that: The real coordinate position of the target human body at the t moment manually input by the user is combined with the predicted state information to obtain the motion state information of the target human body at the t moment, specifically, Calculate the prediction error covariance matrix between the predicted coordinate position and the true coordinate position at time t: （2）， Among them, Q is a 6*6 diagonal matrix, P t-1 is the error covariance matrix between the estimated coordinate position and the true coordinate position at time t-1; Then find the Kalman gain: （3）， Where H is the observation matrix, and , R is a 4*4 diagonal matrix; Finally, calculate the motion state information of the target human body at time t: （4）， Among them, Zt is the real coordinate position of the target body at the tth moment, which is equal to , is the coordinate of the lower midpoint of the real human body rectangular frame detected at the tth moment, The target human body position manually input by the user at time t; Update the error covariance matrix between the estimated coordinate position and the true coordinate position at time t: （5） Where I is the identity matrix.

8. The method of using machine vision to assist manual light tracking according to claim 4, characterized in that: The estimated coordinates of the lower midpoint of the human body rectangular frame in the estimated coordinate position are combined with the camera's position and internal parameters to calculate the spatial coordinate position of the target human body on the stage, and sent to the follow-up light for tracking.

9. According to the method of manual light tracking assisted by machine vision as described in claim 1, the image of the target human body taken by the camera is displayed on the display screen, and the user manually inputs the target human body position at the tth moment as the real coordinate position on the display screen through the cursor.

10. The method of using machine vision to assist manual light tracking according to claim 9, characterized in that: The cursor position is controlled by the user through a mouse, a joystick or a touch screen, and the cursor position at the tth moment is used as the target human body position.

11. A lighting control system, characterized in that: The method of using machine vision to assist manual light chasing as described in any one of claims 1 to 9 comprises a camera with a known posture for capturing images containing a target human body, an input device for a user to manually input the position of the target human body at moment t, and a server and a light chasing light. The server calculates predicted state information of the target human body at moment t including a predicted coordinate position based on the motion state information of the target human body at moment t-1, and combines the real coordinate position with the predicted state information to obtain the motion state information of the target human body at moment t, and calculates the spatial coordinate position of the target human body on the stage by combining the estimated coordinate position with the posture and internal parameters of the camera, and the light chasing light tracks the target human body according to the spatial coordinate position.

Images