Sports event multi-stand collaborative live broadcast method and system based on edge calculation

Abstract

The invention discloses a sports event multi-stand collaborative live broadcast method and system based on edge calculation, and relates to the technical field of digital media, and the method comprises the steps: carrying out the estimation of a sports subject of a sports event; analyzing to obtain the motion condition of the motion subject; acquiring a control habit of the camera; analyzing to obtain at least one deployment position of the camera; performing live broadcast blind area identification on the host position, and matching at least one auxiliary position for the host position; forming a machine position switching mechanism for the host machine position, and forming a live broadcast auxiliary mechanism for the auxiliary machine position; and during live broadcast, the machine position switching mechanism is used for performing live broadcast switching of the host position, and the live broadcast auxiliary mechanism is used for calling the auxiliary machine position corresponding to the host position for live broadcast assistance. Through setting the motion analysis module, the photographing habit acquisition module, the position acquisition module, the position deployment module and the mechanism formation module, the camera position with a better effect is selected for live broadcast, errors caused by manual switching are reduced, and thus the live broadcast effect is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of digital media, and specifically to a multi-camera collaborative live broadcast method and system for sports events based on edge computing. Background Art

[0002] Multi-camera collaborative live broadcast is a technical solution that captures live scenes from different angles through multiple cameras, and through a director system, it can be switched and integrated in real time to finally output high-quality live broadcast content. It is widely used in scenarios such as conferences, events, performances, and e-commerce live broadcasts, and can enhance the visual experience and interactivity of the audience. It is particularly widely used in sports events.

[0003] However, in sports events, the live broadcast of multiple cameras is usually based on experience for position setting, without exhausting all possible camera positions to obtain better camera settings, and the camera switching is also based on experience, but the camera switching is basically operated by one person. Therefore, the live broadcast of multiple cameras places a large load on people, is prone to errors, and affects the live broadcast effect. Summary of the Invention

[0004] To solve the above technical problems, a multi-camera collaborative live broadcast method and system for sports events based on edge computing are provided, and this technical solution solves the problems raised in the above background art.

[0005] To achieve the above object, the technical solution adopted by the present invention is as follows: A multi-camera collaborative live broadcast method for sports events based on edge computing, including: Obtain the type of movement of the sports event as the target type, and based on the target type, estimate the moving object of the sports event; Based on the moving object of the sports event, analyze the movement situation of the moving object; Based on big data, obtain the control habits of the camera; Based on the control habits of the camera and the movement situation of the moving object, analyze at least one deployment position of the camera, and set a camera at each deployment position; Select at least one position from at least one deployment position as the main camera position respectively, identify the live broadcast blind area of the main camera position, and based on the live broadcast blind area of the main camera position, match at least one auxiliary camera position for the main camera position; Form a camera position switching mechanism for the main camera position and a live broadcast assistance mechanism for the auxiliary camera position; During the live broadcast, use the camera position switching mechanism to perform the live broadcast switching of the main camera position, and use the live broadcast assistance mechanism to call the auxiliary camera position corresponding to the main camera position for live broadcast assistance.

[0006] Preferably, the estimating the moving object of the sports event based on the target type includes the following steps: Based on the target type, analyze the main objects participating in the sports event. The main objects are athletes or balls. When the picture of the sports event only contains athletes, the athletes are regarded as the moving main body of the sports event. When the picture of the sports event contains both athletes and balls, the athletes and balls are regarded as the moving main body of the sports event.

[0007] Preferably, the steps of analyzing the motion situation of the moving main body based on the moving main body of the sports event include the following: Based on big data, obtain at least one motion point where the moving main body appears in at least one historical sports event; Obtain the coverage range of at least one motion point that appears in the same historical sports event as the feature range, and associate the feature range with the historical sports event; Scale the event venue and the corresponding feature range in the historical sports event to obtain a reference event venue and a reference feature range, satisfying that the size of the reference event venue is a preset size, and the preset size is the size of the event venue in any historical sports event; Overlap the reference feature ranges, satisfying that the reference event venues corresponding to the reference feature ranges all coincide; Take the area covered by the overlapped reference feature ranges as the feature area, and take the feature area as the motion situation of the moving main body.

[0008] Preferably, the steps of obtaining the control habit of the camera based on big data include the following: Based on historical data, obtain the maximum value of the rotational angular velocity of the camera controlled by the cameraman as the characteristic angular velocity; Based on historical data, obtain at least one imaging ratio of the picture of the moving main body in the camera, and based on user evaluation, obtain the reasonable imaging ratio range of the picture of the moving main body in the camera; Take the characteristic angular velocity and the reasonable imaging ratio range as the control habit of the camera.

[0009] Preferably, the steps of analyzing at least one deployment position of the camera based on the control habit of the camera and the motion situation of the moving main body include the following: Form a target area in the event venue of the current sports event, where the relative position of the target area to the event venue of the current sports event is the same as the relative position of the feature area to the reference event venue, and the proportional relationship between the target area and the event venue of the current sports event is the same as the proportional relationship between the feature area and the reference event venue; Evenly take at least one sampling point in the environment where the event venue of the current sports event is located, and randomly combine several of the sampling points to obtain at least one sampling point combination; Set the perspective of the camera at the sampling point pointing to the center of the event venue in the current sports event as the fixed perspective; Obtain the focal length adjustment range of the camera at the sampling point, satisfying that when the focal length is within the focal length adjustment range, the proportion of the moving object in the picture of the camera at the sampling point belongs to the reasonable imaging proportion range; Under the fixed perspective, obtain the maximum field of view range of the camera at the sampling point with the focal length adjusted within the focal length adjustment range as the fixed field of view range; When the sum of the fixed field of view ranges of the cameras at the sampling points in the sampling point combination includes the target area and the fixed field of view ranges of the cameras at the sampling points in the sampling point combination come from the front, back, left, and right directions of the event venue in the current sports event, then regard the sampling point combination as the preliminary sampling point combination; Regard the sampling points in the preliminary sampling point combination with the fewest number of sampling points as the deployment positions.

[0010] Preferably, the steps of screening at least one position from at least one deployment position to be used as the main camera position respectively include: Classify the deployment positions according to the front, back, left, and right directions of the event venue in the current sports event, and obtain the deployment positions in the same direction, which are summarized into a deployment position set; Regard the coincidence ratio between the fixed field of view range at the deployment position and the target area as the characteristic ratio; Regard the deployment position with the largest characteristic ratio in the deployment position set as the main camera position.

[0011] Preferably, the steps of identifying the live blind area of the main camera position and matching at least one auxiliary camera position based on the live blind area of the main camera position include: Match the points in the target area to the main camera position closest to them, and summarize the points in the target area matched to the main camera position to obtain the main camera position recognition area; Uniformly set at least one recognition point on the main camera position recognition area, and set a cube on the recognition point, and the side length of the cube is one-tenth of the distance between adjacent recognition points; Regard the faces of the cube that cannot be captured by the main camera position as the live blind area; Among the deployment positions other than the main camera position, obtain the deployment positions that can capture the live blind area as the auxiliary camera positions, and perform duplicate removal processing on at least one auxiliary camera position.

[0012] Preferably, the steps of forming a camera position switching mechanism for the main camera position include: When the moving object includes a ball, then regard the moving object as the ball, the current ball-touching moving object, and the last ball-touching moving object as the target moving object; When there is no ball in the moving objects, all the moving objects are regarded as target moving objects; The host position for live broadcast is used as the characteristic host position; In the live broadcast screen of the characteristic host position, according to the movement of the target moving objects in each frame, the moving speed of the target moving objects is obtained. The maximum value of the moving speed of the target moving objects is used as the critical value, and the target moving object with the maximum moving speed is used as the calibrated moving object; The center of the event venue in the current sports event is used as the target point, and at least one sampling point is evenly selected in the event venue and its surrounding environment in the current sports event; The connection line from the characteristic host position to the target point is obtained in advance as the first connection line, and the connection line from the sampling point to the target point is obtained as the second connection line; The lengths of the first connection line and the second connection line are obtained in advance, and the included angle between the first connection line and the second connection line is obtained; The sampling point closest to the calibrated moving object is used as the target sampling point; Using the cosine theorem and the lengths of the first connection line and the second connection line of the target sampling point, the distance between the target sampling point and the characteristic host position is calculated as the target distance; The target distance is multiplied by the characteristic angular velocity to obtain the target speed. When the target speed is less than the critical value, the live broadcast of the characteristic host position is switched, and the live broadcast screen is switched to the host position including the calibrated moving object. The switching moment is the moment when the calibrated moving object disappears in the screen of the characteristic host position.

[0013] Preferably, the formation of the live broadcast assistance mechanism for the auxiliary camera positions includes the following steps: From the perspective of the characteristic host position, calculate the overlapping ratio between the target moving objects as the first occlusion ratio, and take the average value of the first occlusion ratios of all pairwise combinations of the target moving objects to obtain the first verification value; From the perspective of the auxiliary camera position corresponding to the characteristic host position, calculate the overlapping ratio between the target moving objects as the second occlusion ratio, and take the average value of the second occlusion ratios of all pairwise combinations of the target moving objects to obtain the second verification value; When the first verification value is greater than the average value of at least one second verification value, the screen is switched to the auxiliary camera position with the smallest second verification value for live broadcast. When the first verification value is not greater than the average value of at least one second verification value, the live broadcast is switched back to the characteristic host position.

[0014] The multi-camera collaborative live broadcast system for sports events based on edge computing is used to implement the above multi-camera collaborative live broadcast method for sports events based on edge computing, including: A data acquisition module, which acquires the type of sport in a sports event as the target type, and estimates the moving entity of the sports event based on the target type. A motion analysis module, which analyzes the motion situation of the moving entity based on the moving entity of the sports event. A shooting habit acquisition module, which acquires the control habit of the camera based on big data. A position acquisition module, which analyzes at least one deployment position of the camera based on the control habit of the camera and the motion situation of the moving entity, and sets a camera at each deployment position. A position deployment module, which screens out at least one position from at least one deployment position as the main camera positions respectively, identifies the live blind spots of the main camera positions, and matches at least one auxiliary camera position for the main camera positions based on the live blind spots of the main camera positions. A mechanism formation module, which forms a camera position switching mechanism for the main camera positions and a live broadcast assistance mechanism for the auxiliary camera positions. A live broadcast setting module, which uses the camera position switching mechanism to perform live broadcast switching of the main camera positions during live broadcast, and uses the live broadcast assistance mechanism to call the auxiliary camera positions corresponding to the main camera positions for live broadcast assistance.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: By setting a motion analysis module, a shooting habit acquisition module, a position acquisition module, a position deployment module and a mechanism formation module, the live broadcast limit of the camera is estimated according to the habit of the photographer controlling the camera, and based on the estimation result, the effects of all possible combinations of live broadcast camera positions are analyzed, so as to select a camera position with better effect for live broadcast, and set the main camera positions and auxiliary camera positions according to the positions of the camera positions. At the same time, a camera position switching mechanism is also formed, so that during live broadcast, multiple camera positions can be identified by an autonomous recognition method, and errors caused by manual switching during high-intensity live broadcast can be reduced, thus ensuring the live broadcast effect. Description of the Drawings

[0016] Figure 1 It is a schematic flow chart of the multi-camera collaborative live broadcast method for sports events based on edge computing of the present invention; Figure 2 It is a schematic flow chart of analyzing the motion situation of the moving entity based on the moving entity of the sports event of the present invention; Figure 3 It is a schematic flow chart of acquiring the control habit of the camera based on big data of the present invention; Figure 4Flow chart for analyzing at least one deployment position of a camera based on the control habits of the camera and the movement of a moving object according to the present invention; Figure 5 Flow chart for screening at least one position from at least one deployment position and respectively using it as a main camera position according to the present invention; Figure 6 Flow chart for identifying live blind spots for a main camera position and matching at least one auxiliary camera position for the main camera position based on the live blind spots of the main camera position according to the present invention; Figure 7 Flow chart for forming a camera position switching mechanism for a main camera position according to the present invention; Figure 8 Flow chart for forming a live broadcast assistance mechanism for an auxiliary camera position according to the present invention. Detailed implementation mode

[0017] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments in the following description are only examples, and those skilled in the art can think of other obvious variations.

[0018] Refer to Figure 1 As shown, the multi-camera collaborative live broadcast method for sports events based on edge computing includes: Obtain the type of movement of a sports event as a target type, and estimate the moving object of the sports event based on the target type; Analyze the movement of the moving object based on the moving object of the sports event; Obtain the control habits of the camera based on big data; Analyze at least one deployment position of the camera based on the control habits of the camera and the movement of the moving object, and set a camera at each deployment position; Screen at least one position from at least one deployment position and respectively use it as a main camera position, identify the live blind spots of the main camera position, and match at least one auxiliary camera position for the main camera position based on the live blind spots of the main camera position; Form a camera position switching mechanism for the main camera position and a live broadcast assistance mechanism for the auxiliary camera position; During live broadcast, use the camera position switching mechanism to perform live broadcast switching of the main camera position, and use the live broadcast assistance mechanism to call the auxiliary camera position corresponding to the main camera position for live broadcast assistance.

[0019] In this solution, the camera position of the live broadcast is mainly set. Because during the live broadcast, there may be situations where the players block each other or the ball is blocked by the players, so the live broadcast perspective needs to be changed to obtain a better viewing experience from other perspectives. In addition, since the ball moves quickly, its movement speed may be very fast. For example, a fast break in a basketball game may exceed the camera lens movement speed, resulting in out-of-frame situations. Therefore, in order to avoid this situation, it is necessary to ensure that the field of view of all camera positions covers the possible moving positions of the ball and the players. Therefore, the camera position can be switched to ensure that the ball and the players are in the live broadcast screen. It should be noted that although the ball and the athletes can be kept in the field of view by expanding the field of view, the viewing experience will be poor if the ball and the athletes occupy too small a proportion of the screen. Therefore, in order to ensure the viewing experience, it is necessary to ensure that the ball and the athletes occupy a reasonable proportion of the screen. This will result in the photographer being unable to track the target when the speed suddenly changes, which is very common in ball sports, so screen switching is required; The live broadcast of this scheme is different from the usual live broadcast. This scheme uses a computer to autonomously switch the images of different camera positions. It is common sense that when the ball and the athletes occupy a relatively large proportion in the picture, the viewing is clearer and the viewing experience is better, but this will result in a smaller field of view. Therefore, more camera positions need to be set, resulting in a large number of images that may need to be switched, and manual operation is not competent. Therefore, it reduces the number of camera positions by enlarging the field of view, but this will result in room for improvement in the live broadcast effect. This is very common in basketball games, many of which use half-court coverage for live broadcasts. The athletes occupy a very small proportion in the picture, and their physical confrontation and dribbling techniques are difficult to reflect in the picture. Therefore, in this scheme, relatively more camera positions are set, and the live broadcast does not use a large-area coverage method for the stadium. This will result in a large number of images available for switching. Therefore, a series of mechanisms are set up later for autonomous switching to ensure that there will be no errors in the live broadcast; In this scheme, only the games in the field are considered, and the games in open spaces such as racing and marathon are not considered. There are athletes in every game, but not all games have balls. Therefore, this will be distinguished in the subsequent processing.

[0020] Based on the target type, estimating the sports subject of the sports event includes the following steps: Based on the target type, the main objects participating in the sports event are analyzed. The main objects are athletes or balls. When the picture of the sports event only contains athletes, the athletes are regarded as the sports subjects of the sports event. When the picture of the sports event contains athletes and balls, the athletes and the ball are regarded as the sports subjects of the sports event.

[0021] Reference Figure 2As shown in the figure, the steps for analyzing the motion of a moving object based on a sports event are as follows: Based on big data, obtain at least one motion point where the moving object appears in at least one historical sports event; Obtain the coverage range of at least one motion point that appears in the same historical sports event as the feature range, and associate the feature range with the historical sports event; Scale the event venue and the corresponding feature range in the historical sports event to obtain a reference event venue and a reference feature range, where the size of the reference event venue meets the preset size, and the preset size is the size of the event venue in any historical sports event; Overlap the reference feature ranges to ensure that the reference event venues corresponding to the reference feature ranges coincide; Take the area covered by the overlapped reference feature ranges as the feature area, and use the feature area as the motion situation of the moving object.

[0022] For example, the event venue is a basketball court, a table tennis table, etc. However, the live broadcast images are not limited to these, but also include the surrounding environment. In different situations, the motion ranges of athletes and balls are different, and it is necessary to capture their possible motion positions, so as to set the subsequent camera positions according to this situation.

[0023] Refer to Figure 3 As shown in the figure, the steps for obtaining the control habits of the camera based on big data are as follows: Based on historical data, obtain the maximum value of the rotational angular velocity at which the cameraman controls the camera as the characteristic angular velocity; Based on historical data, obtain at least one imaging ratio of the moving object in the camera image, and obtain the reasonable imaging ratio range of the moving object in the camera image based on user evaluation; Take the characteristic angular velocity and the reasonable imaging ratio range as the control habits of the camera.

[0024] On the one hand, it is necessary to consider how large the moving object appears in the image for a better viewing experience. On the other hand, it is also necessary to consider whether the cameraman can keep up with the speed of the moving object by rotating the perspective. This mainly deals with the sudden acceleration of the ball in ball games. Because the field of view of the camera is small, if the target cannot be tracked, the target will go out of the frame.

[0025] Refer to Figure 4 As shown in the figure, the steps for analyzing at least one deployment position of the camera based on the control habits of the camera and the motion situation of the moving object are as follows: A target area is formed in the event venue of the current sports event. The relative position of the target area to the event venue of the current sports event is the same as the relative position of the feature area to the reference event venue, and the proportional relationship between the target area and the event venue of the current sports event is the same as the proportional relationship between the feature area and the reference event venue; At least one sampling point is evenly selected in the environment where the event venue of the current sports event is located, and several of the sampling points are randomly combined to obtain at least one sampling point combination; The perspective of pointing the camera at the center of the event venue of the current sports event at the sampling point is used as the fixed perspective; Obtain the focal length adjustment range of the camera at the sampling point, satisfying that when the focal length is within the focal length adjustment range, the proportion of the moving object in the picture of the camera at the sampling point belongs to the reasonable imaging proportion range; Under the fixed perspective, obtain the maximum field of view range of the camera at the sampling point where the focal length is adjusted within the focal length adjustment range, as the fixed field of view range; When the sum of the fixed field of view ranges of the cameras at the sampling points in the sampling point combination includes the target area and the fixed field of view ranges of the cameras at the sampling points in the sampling point combination come from the four directions of the front, back, left, and right of the event venue of the current sports event, then the sampling point combination is used as the preliminary sampling point combination; The sampling points in the preliminary sampling point combination with the fewest number of sampling points are used as the deployment positions.

[0026] When setting the deployment positions, it is necessary to ensure that it can broadcast the target area from multiple angles, so that there can be room for switching live pictures later. Here, only the four directions of the front, back, left, and right need to be considered. At the same time, it is necessary to ensure that it covers all positions of the target area, so that the target can always be guaranteed to be in the picture through switching.

[0027] Refer to Figure 5 As shown, screening at least one position from at least one deployment position to be used as the main camera position respectively includes the following steps: Classify the deployment positions according to the four directions of the front, back, left, and right of the event venue of the current sports event, and the deployment positions located in the same direction are summarized into a deployment position set; The overlapping ratio of the fixed field of view range at the deployment position to the target area is used as the characteristic ratio; The deployment position with the largest characteristic ratio in the deployment position set is used as the main camera position.

[0028] When conducting a live broadcast, screening is carried out according to the distribution of deployment locations. It is necessary to select a location with a better view as the main camera position. The one with a better view effect must have the largest overlapping ratio with the target area. Because the larger the overlapping view, the less adjustment the cameraman needs to make to the camera. Therefore, the possibility of misoperation is smaller.

[0029] Refer to Figure 6 As shown, for identifying the live broadcast blind area of the main camera position, based on the live broadcast blind area of the main camera position, matching at least one auxiliary camera position for the main camera position includes the following steps: Match the points in the target area to the main camera position with the shortest distance to them, and summarize the points in the target area that are matched to the main camera position to obtain the main camera position identification area; Uniformly set at least one identification point on the main camera position identification area, and set a cube on the identification point. The side length of the cube is one-tenth of the distance between adjacent identification points; Regard the faces of the cube that cannot be captured by the main camera position as the live broadcast blind area; Among the deployment locations other than the main camera position, obtain the deployment locations that can capture the live broadcast blind area as auxiliary camera positions, and perform deduplication processing on at least one auxiliary camera position.

[0030] The main camera position broadcasts the event from only one direction. Therefore, there may be too many occlusions, which will have a greater impact on the live broadcast effect. Use a cube to simulate the possible occlusion situation of the moving object, so as to obtain a camera position that can supplement the blind area of the main camera position as an auxiliary camera position.

[0031] Refer to Figure 7 As shown, forming a camera position switching mechanism for the main camera position includes the following steps: When the moving object contains a ball, regard the moving object as the ball, the currently ball-touching moving object, and the last ball-touching moving object as the target moving objects; When the moving object does not contain a ball, regard all moving objects as the target moving objects; Regard the main camera position for live broadcast as the feature main camera position; In the live broadcast screen of the feature main camera position, according to the movement situation of the target moving object in each frame of the screen, obtain the moving speed of the target moving object. Take the maximum value of the moving speed of the target moving object as the critical value, and regard the target moving object with the maximum moving speed as the calibrated moving object; Regard the center of the event venue in the current sports event as the target point, and uniformly take at least one sampling point in the event venue and its surrounding environment in the current sports event; Pre-obtain the connection line from the feature main camera position to the target point as the first connection line, and obtain the connection line from the sampling point to the target point as the second connection line; Pre-acquire the lengths of the first connection line and the second connection line, and obtain the included angle between the first connection line and the second connection line; Take the sampling point closest to the calibrated moving object as the target sampling point; Use the cosine theorem and the lengths of the first connection line and the second connection line of the target sampling point to calculate the distance between the target sampling point and the feature host position as the target distance; Multiply the target distance by the feature angular velocity to obtain the target speed. When the target speed is less than the critical value, the live broadcast of the feature host position is switched, and the live broadcast screen is switched to the host position including the calibrated moving object. The switching moment is the moment when the calibrated moving object disappears from the screen of the feature host position.

[0032] The switching of the host position is mainly to replace the screen of the feature host position with the screen of another host position when the speed of the moving object is suddenly very fast and there is a possibility of going out of the picture. Whether it will go out of the picture mainly depends on whether the speed of the cameraman tracking the target can keep up with the critical value. And the angular velocity of the cameraman controlling the rotation of the camera has an upper limit, which is the feature angular velocity. Then the tracking speed at the calibrated moving object is equal to the target speed. As long as its speed is faster than the critical value, the cameraman can track any target in the picture according to the actual situation, so as to ensure that the required target will not go out of the picture. Otherwise, it may not be able to track, resulting in the required target going out of the picture; The cosine theorem is as follows: , where c is the target distance, a is the length of the first connection line, b is the length of the second connection line of the target sampling point, and C is the included angle between the first connection line and the second connection line of the target sampling point.

[0033] Refer to Figure 8 As shown, the formation of a live broadcast assistance mechanism for the auxiliary camera positions includes the following steps: From the perspective of the feature host position, calculate the overlapping ratio between the target moving objects as the first occlusion ratio, and take the average value of the first occlusion ratios of all pairwise combinations of the target moving objects to obtain the first verification value; From the perspective of the auxiliary camera position corresponding to the feature host position, calculate the overlapping ratio between the target moving objects as the second occlusion ratio, and take the average value of the second occlusion ratios of all pairwise combinations of the target moving objects to obtain the second verification value; When the first verification value is greater than the average value of at least one second verification value, the screen is switched to the auxiliary camera position with the smallest second verification value for live broadcast. When the first verification value is not greater than the average value of at least one second verification value, the live broadcast is switched back to the feature host position.

[0034] When the situation where the feature host position is blocked is not greater than the average value of the auxiliary camera positions, it indicates that its live broadcast effect is acceptable and no adjustment is required. However, when the blocking situation is too large, it indicates that there may be a certain impact, so it is necessary to temporarily replace it with the picture of the auxiliary camera position.

[0035] The multi-camera collaborative live broadcast system for sports events based on edge computing is used to implement the above-mentioned multi-camera collaborative live broadcast method for sports events based on edge computing, and includes: A data acquisition module, which acquires the type of movement of the sports event as the target type, and estimates the moving object of the sports event based on the target type; A motion analysis module, which analyzes the motion situation of the moving object based on the moving object of the sports event; A shooting habit acquisition module, which acquires the control habit of the camera based on big data; A position acquisition module, which analyzes at least one deployment position of the camera based on the control habit of the camera and the motion situation of the moving object, and a camera is set at each deployment position; A position deployment module, which screens out at least one position from at least one deployment position as the host position respectively, identifies the live broadcast blind area of the host position, and matches at least one auxiliary camera position for the host position based on the live broadcast blind area of the host position; A mechanism formation module, which forms a camera position switching mechanism for the host position and a live broadcast assistance mechanism for the auxiliary camera position; A live broadcast setting module, which uses the camera position switching mechanism to perform live broadcast switching of the host position during live broadcast, and uses the live broadcast assistance mechanism to call the auxiliary camera position corresponding to the host position for live broadcast assistance.

[0036] Furthermore, this solution also proposes a storage medium, on which a computer-readable program is stored. When the computer-readable program is called, it executes the above-mentioned multi-camera collaborative live broadcast method for sports events based on edge computing.

[0037] It can be understood that the storage medium can be a magnetic medium, such as a floppy disk, a hard disk, a magnetic tape; an optical medium such as a DVD; or a semiconductor medium such as a solid state disk (SSD), etc.

[0038] In summary, the advantages of the present invention are as follows: By setting up a motion analysis module, a photography habit acquisition module, a position acquisition module, a position deployment module, and a mechanism formation module, the live broadcast limit of the camera is estimated according to the habits of the photographer controlling the camera. Based on the estimation results, the effects of all possible combinations of live broadcast camera positions are analyzed, so as to select a camera position with better effects for live broadcast, and the main camera position and auxiliary camera positions are set according to the positions of the camera positions. At the same time, a mechanism for camera position switching is formed, so that during the live broadcast, multiple camera positions can be recognized by means of autonomous recognition, and errors caused by manual switching can be reduced during the high-intensity live broadcast process, thus ensuring the live broadcast effect.

[0039] The foregoing has shown and described the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. What is described in the above embodiments and the specification is only the principle of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection required by the present invention is defined by the appended claims and their equivalents.

Claims

1. A multi-camera collaborative live broadcast method for sports events based on edge computing, characterized in that: include: Obtaining the sport type of the sports event as a target type, and estimating the sport subject of the sports event based on the target type; Based on the sports subjects of sports events, the sports conditions of the sports subjects are analyzed; Based on big data, obtain the control habits of the camera; Based on the control habits of the camera and the movement of the moving subject, at least one deployment position of the camera is obtained by analysis, and a camera is set at each deployment position; At least one position is selected from at least one deployment position as a main position, a live broadcast blind area is identified for the main position, and at least one auxiliary position is matched for the main position based on the live broadcast blind area of ​​the main position; A camera switching mechanism is formed for the main camera, and a live broadcast auxiliary mechanism is formed for the auxiliary camera; During live broadcast, use the camera switching mechanism to switch the host position to live broadcast, and use the live broadcast auxiliary mechanism to call the auxiliary camera corresponding to the host position to assist in live broadcast.

2. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 1, characterized in that: The method of estimating the sports subject of a sports event based on the target type comprises the following steps: Based on the target type, the main objects participating in the sports event are analyzed. The main objects are athletes or balls. When the picture of the sports event only contains athletes, the athletes are regarded as the sports subjects of the sports event. When the picture of the sports event contains athletes and balls, the athletes and the ball are regarded as the sports subjects of the sports event.

3. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 2, characterized in that: The analyzing of the sports subject based on the sports event to obtain the sports condition of the sports subject comprises the following steps: Based on big data, obtaining at least one sports point where a sports subject appears in at least one historical sports event; Obtaining a coverage range of at least one sports point appearing in the same historical sports event as a feature range, and matching the feature range with the historical sports event; Scaling the event venues and corresponding feature ranges in historical sports events to obtain a benchmark event venue and a benchmark feature range, where the size of the benchmark event venue is a preset size, and the preset size is the size of the event venue in any historical sports event; Overlap the benchmark feature ranges to ensure that the benchmark event venues corresponding to the benchmark feature ranges are overlapped; The area covered by the overlapped reference feature range lock is taken as the feature area, and the feature area is taken as the movement situation of the moving subject.

4. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 3 is characterized in that: The method of obtaining the control habits of the camera based on big data includes the following steps: Based on the historical data, the maximum value of the rotation angular velocity of the camera controlled by the camera operator is obtained as the characteristic angular velocity; Based on historical data, at least one imaging ratio of the moving subject in the camera's picture is obtained, and based on user evaluation, a reasonable imaging ratio range of the moving subject in the camera's picture is obtained; The characteristic angular velocity and reasonable imaging ratio range are used as the control habits of the camera.

5. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 4 is characterized in that: The step of analyzing and obtaining at least one deployment position of the camera based on the control habits of the camera and the movement of the moving subject comprises the following steps: A target area is formed in the event venue of the current sports event, the relative position of the target area to the event venue of the current sports event is consistent with the relative position of the characteristic area to the benchmark event venue, and the proportional relationship between the target area and the event venue of the current sports event is consistent with the proportional relationship between the characteristic area and the benchmark event venue; Uniformly select at least one sampling point in the environment where the event venue of the current sports event is located, and randomly combine several of the sampling points to obtain at least one sampling point combination; The camera set at the sampling point points to the center of the event venue in the current sports event as a fixed angle of view; Obtaining a focal length adjustment range of the camera at the sampling point, which satisfies the following conditions: when the focal length is within the focal length adjustment range, the proportion of the moving subject in the image of the camera at the sampling point belongs to a reasonable imaging proportion range; Under a fixed viewing angle, obtaining a maximum field of view of the camera at a sampling point where the focal length is adjusted within a focal length adjustment range as a fixed field of view; When the sum of the fixed field of view of the cameras at the sampling points in the sampling point combination includes the target area and the fixed field of view of the cameras at the sampling points in the sampling point combination is from the front, back, left, right and four directions of the event venue in the current sports event, the sampling point combination is used as a preliminary sampling point combination; The sampling points in the combination of prepared sampling points that contains the least number of sampling points are selected as deployment locations.

6. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 5, characterized in that: The step of selecting at least one location from at least one deployment location as a host location comprises the following steps: According to the four directions of the front, back, left, and right of the event venue in the current sports event, the deployment positions are classified, and the deployment positions located in the same direction are summarized into a deployment position set; The overlap ratio between the fixed field of view at the deployment position and the target area is used as the feature ratio; The deployment location with the largest feature ratio in the deployment location set is taken as the host location.

7. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 6, characterized in that: The method of identifying the live broadcast blind area of ​​the main machine position and matching at least one auxiliary machine position for the main machine position based on the live broadcast blind area of ​​the main machine position comprises the following steps: Match the points in the target area to the host position closest to it, summarize the points in the target area matched to the host position, and obtain the host position identification area; At least one identification point is evenly arranged on the host position identification area, and a cube is arranged on the identification point, and the side length of the cube is one tenth of the distance between adjacent identification points; The faces of the cube that cannot be captured by the host are used as live broadcast blind spots; Among the deployment positions other than the main camera position, a deployment position that can capture the blind spot of the live broadcast is obtained as an auxiliary camera position, and deduplication processing is performed on at least one auxiliary camera position.

8. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 7, characterized in that: The forming of a machine position switching mechanism for the host position comprises the following steps: When the moving subject includes a ball, the moving subject that is the ball, the moving subject that is currently touching the ball, and the moving subject that last touched the ball are taken as the target moving subject; When the moving subject does not contain the ball, all moving subjects are regarded as target moving subjects; The host position for live broadcast is used as the featured host position; In the live broadcast picture of the characteristic host position, the moving speed of the target moving subject is obtained according to the moving situation of the target moving subject in each frame, the maximum moving speed of the target moving subject is used as the critical value, and the target moving subject with the largest moving speed is used as the calibration moving subject; The center of the event venue in the current sports event is taken as the target point, and at least one sampling point is evenly selected in the event venue in the current sports event and its surrounding environment; Pre-acquire a line from the characteristic host position to the target point as a first line, and acquire a line from the sampling point to the target point as a second line; Pre-acquire the length of the first connecting line and the second connecting line, and acquire the angle between the first connecting line and the second connecting line; The sampling point closest to the calibrated moving subject is taken as the target sampling point; Using the law of cosines, the length of the first connecting line and the second connecting line of the target sampling point, the distance between the target sampling point and the characteristic host position is calculated as the target distance; The target distance is multiplied by the characteristic angular velocity to obtain the target speed. When the target speed is less than the critical value, the live broadcast is switched to the host position containing the calibrated moving subject. The switching moment is the moment when the calibrated moving subject disappears from the picture of the characteristic host position.

9. The method for multi-camera collaborative live broadcast of sports events based on edge computing according to claim 8, characterized in that: The method of forming a live broadcast auxiliary mechanism for the auxiliary camera position includes the following steps: Under the perspective of the characteristic host position, the overlap ratio between the target moving bodies is calculated as the first occlusion ratio, and the first occlusion ratios of all pairwise combinations of the target moving bodies are averaged to obtain a first verification value; Under the perspective of the auxiliary camera position corresponding to the characteristic main camera position, the overlap ratio between the target moving subjects and the target moving subjects is calculated as the second occlusion ratio, and the second occlusion ratios of all pairwise combinations of the target moving subjects are averaged to obtain a second verification value; When the first check value is greater than the average of at least one second check value, the screen is switched to the auxiliary camera with the smallest second check value for live broadcast; when the first check value is not greater than the average of at least one second check value, the screen is switched back to the characteristic host camera for live broadcast.

10. A sports event multi-camera collaborative live broadcast system based on edge computing, used to implement a sports event multi-camera collaborative live broadcast method based on edge computing as described in any one of claims 1 to 9, characterized in that: include: A data acquisition module, wherein the data acquisition module acquires the sport type of the sports event as a target type, and estimates the sport subject of the sports event based on the target type; A motion analysis module, wherein the motion analysis module analyzes the motion conditions of the motion subjects based on the motion subjects of the sports events; A photography habit acquisition module, wherein the photography habit acquisition module acquires the control habits of the camera based on big data; A position acquisition module, wherein the position acquisition module analyzes and obtains at least one deployment position of the camera based on the control habits of the camera and the movement of the moving subject, and a camera is arranged at each deployment position; A position deployment module, wherein the position deployment module selects at least one position from at least one deployment position as a main position, identifies a live broadcast blind area for the main position, and matches at least one auxiliary position for the main position based on the live broadcast blind area of ​​the main position; A mechanism forming module, wherein the mechanism forming module forms a camera switching mechanism for the main camera and forms a live broadcast auxiliary mechanism for the auxiliary camera; The live broadcast setting module uses the camera switching mechanism to switch the live broadcast of the host position during the live broadcast, and uses the live broadcast auxiliary mechanism to call the auxiliary camera corresponding to the host position to assist in the live broadcast.

Images