WO2020107480A1 - 图像特征点的评价方法和可移动平台 - Google Patents
图像特征点的评价方法和可移动平台 Download PDFInfo
- Publication number
- WO2020107480A1 WO2020107480A1 PCT/CN2018/118778 CN2018118778W WO2020107480A1 WO 2020107480 A1 WO2020107480 A1 WO 2020107480A1 CN 2018118778 W CN2018118778 W CN 2018118778W WO 2020107480 A1 WO2020107480 A1 WO 2020107480A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- image feature
- feature points
- dimensional coordinates
- image
- images
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Ceased
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
Definitions
- the embodiments of the present invention relate to the field of image processing, and in particular, to an image feature point evaluation method and a movable platform.
- the image feature points of the target are generally extracted from multi-frame images, and the image feature points extracted from the multi-frame images are matched to calculate the three-dimensional coordinates of each image feature point in the world coordinate system After the coordinates are applied to the UAV field, the UAV can track the target according to the three-dimensional coordinates.
- the image feature points are determined according to the uniqueness of the image feature points and the matching degree of matching.
- the uniqueness is that there is no other image feature point similar to the image feature point
- the matching degree is the similarity between the image feature point and the image feature point matched with it on other images.
- the good image feature points determined in the multi-frame images will not actually match exactly when matching, so that the calculated three-dimensional coordinates of the image feature points are inaccurate, resulting in the inability to accurately track Target.
- Embodiments of the present invention provide an image feature point evaluation method and a mobile platform to achieve accurate evaluation of image features with high reliability, so that image features with high reliability can be used to track targets on a mobile platform Object, which can improve the accuracy of tracking the target object.
- an embodiment of the present invention provides an image feature point evaluation method, which is applied to a mobile platform and includes:
- N Acquiring N images acquired by the shooting device under N different camera positions, wherein each image includes an image of a target, and N is an integer greater than or equal to 2;
- the image feature points are evaluated.
- an embodiment of the present invention provides a movable platform, including: a processor and a shooting device;
- the shooting device is used to collect and obtain N images under N different camera positions
- the processor is configured to acquire N images acquired by the shooting device under N different camera positions, wherein each image includes an image of a target, and N is an integer greater than or equal to 2; extract the target Related information of the same image feature points of the images in the N images; based on the related information of the image feature points in the N images, to determine the relative difference of the image feature points; according to the Relative differences are evaluated on the image feature points.
- an embodiment of the present invention provides a computer-readable storage medium that stores a computer program, where the computer program includes at least one piece of code that can be executed by a computer to control The computer executes the image feature point evaluation method described in the first aspect of the present invention.
- an embodiment of the present invention provides a computer program.
- the computer program When the computer program is executed by a computer, it is used to implement the image feature point evaluation method described in the first aspect of the present invention.
- the image feature point evaluation method and the movable platform provided by the embodiments of the present invention, by acquiring N images acquired by the shooting device under N different camera positions, and then extracting the same image feature points of the target object image respectively According to the relevant information in the image, the relative difference of the image feature points is determined according to the relevant information of the image feature points in the N images, and finally, the image feature points are evaluated according to the relative difference. Therefore, the image feature points with high reliability can be accurately evaluated, and the image feature points with high reliability can be used for the mobile platform to track the target, so that the accuracy of tracking the target can be improved.
- FIG. 1 is a schematic architectural diagram of an unmanned aerial system according to an embodiment of the present invention
- FIG. 2 is a flowchart of an image feature point evaluation method according to an embodiment of the present invention.
- FIG. 3 is a schematic structural diagram of a movable platform provided by an embodiment of the present invention.
- a component when a component is said to be “fixed” to another component, it can be directly on another component or it can also exist in a centered component. When a component is considered to be “connected” to another component, it can be directly connected to another component or there can be centered components at the same time.
- the embodiment of the present invention provides an image feature point evaluation method and a movable platform.
- the movable platform may be, for example, a drone, an unmanned boat, an unmanned car, a robot, and the like.
- the drone may be, for example, a rotorcraft (rotorcraft), for example, a multirotor aircraft propelled by a plurality of propulsion devices through air, and the embodiments of the present invention are not limited thereto.
- FIG. 1 is a schematic architectural diagram of an unmanned aerial system according to an embodiment of the present invention.
- a rotary-wing UAV is taken as an example for description.
- the unmanned aerial system 100 may include a drone 110, a display device 130, and a control terminal 140.
- the UAV 110 may include a power system 150, a flight control system 160, a rack, and a gimbal 120 carried on the rack.
- the drone 110 can communicate with the control terminal 140 and the display device 130 wirelessly.
- the rack may include a fuselage and a tripod (also called landing gear).
- the fuselage may include a center frame and one or more arms connected to the center frame, the one or more arms extending radially from the center frame.
- the tripod is connected to the fuselage and is used to support the UAV 110 when it lands.
- the power system 150 may include one or more electronic governors (abbreviated as electric governors) 151, one or more propellers 153, and one or more motors 152 corresponding to the one or more propellers 153, wherein the motor 152 is connected to Between the electronic governor 151 and the propeller 153, the motor 152 and the propeller 153 are disposed on the arm of the drone 110; the electronic governor 151 is used to receive the driving signal generated by the flight control system 160 and provide driving according to the driving signal The current is given to the motor 152 to control the rotation speed of the motor 152. The motor 152 is used to drive the propeller to rotate, thereby providing power for the flight of the drone 110, which enables the drone 110 to achieve one or more degrees of freedom of movement.
- electric governors abbreviated as electric governors
- the drone 110 may rotate about one or more rotation axes.
- the rotation axis may include a roll axis (Roll), a yaw axis (Yaw), and a pitch axis (Pitch).
- the motor 152 may be a DC motor or an AC motor.
- the motor 152 may be a brushless motor or a brush motor.
- the flight control system 160 may include a flight controller 161 and a sensing system 162.
- the sensing system 162 is used to measure the attitude information of the drone, that is, the position information and status information of the drone 110 in space, for example, three-dimensional position, three-dimensional angle, three-dimensional velocity, three-dimensional acceleration, and three-dimensional angular velocity.
- the sensing system 162 may include, for example, at least one of a gyroscope, an ultrasonic sensor, an electronic compass, an inertial measurement unit (Inertial Measurement Unit, IMU), a visual sensor, a global navigation satellite system, and a barometer.
- the global navigation satellite system may be a global positioning system (Global Positioning System, GPS).
- the flight controller 161 is used to control the flight of the drone 110.
- the flight of the drone 110 can be controlled according to the attitude information measured by the sensor system 162. It should be understood that the flight controller 161 may control the drone 110 according to pre-programmed program instructions, or may control the drone 110 by responding to one or more control instructions from the control terminal 140.
- the gimbal 120 may include a motor 122.
- the gimbal is used to carry the shooting device 123.
- the flight controller 161 can control the movement of the gimbal 120 through the motor 122.
- the gimbal 120 may further include a controller for controlling the movement of the gimbal 120 by controlling the motor 122.
- the gimbal 120 may be independent of the drone 110, or may be a part of the drone 110.
- the motor 122 may be a DC motor or an AC motor.
- the motor 122 may be a brushless motor or a brush motor.
- the gimbal can be located at the top of the drone or at the bottom of the drone.
- the shooting device 123 may be, for example, a device for capturing images such as a camera or a video camera.
- the shooting device 123 may communicate with the flight controller and perform shooting under the control of the flight controller.
- the photographing device 123 of this embodiment at least includes a photosensitive element, for example, a complementary metal oxide semiconductor (Complementary Metal Oxide Semiconductor (CMOS) sensor or a charge-coupled device (Charge-coupled Device, CCD) sensor. It can be understood that the shooting device 123 can also be directly fixed on the drone 110, so that the gimbal 120 can be omitted.
- CMOS Complementary Metal Oxide Semiconductor
- CCD charge-coupled Device
- the display device 130 is located on the ground end of the unmanned aerial system 100, can communicate with the drone 110 in a wireless manner, and can be used to display the attitude information of the drone 110.
- the image captured by the photographing device may also be displayed on the display device 130. It should be understood that the display device 130 may be an independent device or may be integrated in the control terminal 140.
- the control terminal 140 is located at the ground end of the unmanned aerial system 100, and can communicate with the drone 110 in a wireless manner for remote manipulation of the drone 110.
- the drone 110 can recognize the target in the image captured by the shooting device 123 to track the target.
- FIG. 2 is a flowchart of an image feature point evaluation method according to an embodiment of the present invention. As shown in FIG. 2, the method of this embodiment can be applied to a mobile platform. The method of this embodiment can include:
- Each image includes an image of the target object, and N is an integer greater than or equal to 2.
- the movable platform is equipped with a shooting device.
- the shooting device can be used to collect images.
- the shooting device can collect N images under N different camera positions. Each image obtained includes the target’s image.
- Image, N is an integer greater than or equal to 2.
- the shooting device acquires the above images in different poses.
- the different poses may be that the shooting device is at a different spatial position, or at a different rotation angle, etc., which is not limited in this embodiment.
- the image feature points are the feature points on the target.
- the image feature points are, for example, M, and M is an integer greater than or equal to 2.
- M is an integer greater than or equal to 2.
- the number of image feature points on each image may be different. For example, when the shooting device shoots the target object at different camera positions, the change in the shooting angle of the shooting device and/or the displacement of the target object causes the target object At least one feature point on the image cannot be collected by the camera, resulting in a different number of image feature points (that is, feature points on the target) on each image.
- the relevant information of the same image feature point when presented on different images is different.
- the relevant information may be the position information of the image feature point on different images.
- the position information may be the actual position of the image feature point on the image.
- the actual two-dimensional coordinates are the position of the feature point on the target object (that is, the image feature point) on the image that can be actually measured after the target object is collected by the shooting device to obtain the image of the target object.
- Two-dimensional coordinates in a dimensional coordinate system It should be noted that the method of setting the two-dimensional coordinate system on each image is the same.
- S203 Determine the relative difference of the image feature points according to the related information of the image feature points in the N images, respectively.
- the image feature point is in all N
- the relative difference in the images is, for example, the degree of difference between the actual two-dimensional coordinates of the image feature points in all N images, and may include variance.
- the image feature points are evaluated according to the relative differences of the image feature points. For example, if the relative difference is small, the image feature point is an image feature point with high reliability, and if the relative difference is large, the image feature point is an image feature point with low reliability.
- the image feature point when the relative difference is 1, the image feature point is considered to be a high-reliability image feature point; when the relative difference is 20, the image feature point is considered to be low-reliability. Image feature points. Then, in this embodiment, the target object can be tracked according to the image feature points with high informability, so as to improve the accuracy of tracking the target object.
- a possible implementation manner of the above S204 is: according to the relative difference of the image feature points and the relative differences of other image feature points, the image Feature points are evaluated.
- the other image feature points may be other M-1 image feature points, or part of the other M-1 image feature points, which is not limited in this embodiment of the present invention.
- a possible implementation manner of evaluating the image feature points is: if the relative difference of the image feature points is The first K image feature points with the smallest relative difference among the relative differences between the image feature points and the other image feature points, the image feature points are evaluated as image feature points with high reliability.
- K is an integer greater than or equal to 1.
- the size of K can be changed according to the complexity of the image, the brightness of the image, and the mode in which the image feature points with high reliability are selected, which is not limited in this embodiment.
- the relative differences of the M image feature points are compared, and the magnitudes of the relative differences of the M image feature points are sorted , And then select the first K image feature points with the smallest relative difference of the image feature points. If the image feature points are included in the first K image feature points with the smallest relative difference of the image feature points, then the image feature point is credible High-resolution image feature points. If the image feature point is not included in the top K image feature points with the smallest relative difference of the image feature points, it means that the image feature point is an image feature point with low reliability.
- another possible implementation manner of the above S204 is: if the relative difference of the image feature points is less than a preset difference, the image feature points are evaluated as image feature points with high reliability.
- the relative difference between the image feature points and the preset difference are compared. If the relative difference of the image feature points is less than the preset difference, it means that the image feature point is highly reliable If the relative difference of the image feature points is greater than or equal to the preset difference, it means that the image feature point is an image feature point with low reliability.
- another possible implementation manner of the above S204 is: if the relative difference of the image feature points is less than a preset difference, and the relative difference of the image feature points is the image feature points and the other Among the first K image feature points with the smallest relative difference among the relative differences of the image feature points, the image feature points are evaluated as image feature points with high reliability.
- the relative difference between the M image feature points and the preset difference are first compared to obtain at least that the relative difference is less than the preset difference An image feature point. Then, the relative difference of the at least one image feature point is compared to obtain the first K image feature points with the smallest relative difference of the image feature points, and the first K image feature points are image feature points with high reliability.
- the image feature points of the differences are all image feature points with high reliability.
- the image feature point evaluation method provided in this embodiment obtains N images acquired by the shooting device under N different camera positions, and then extracts relevant information of the same image feature points of the target object image in the N images, respectively Based on the relevant information of the image feature points in the N images, the relative difference of the image feature points is determined, and finally the image feature points are evaluated according to the relative difference. Therefore, the image feature points with high reliability can be accurately evaluated, and the image feature points with high reliability can be used for the mobile platform to track the target, so that the accuracy of tracking the target can be improved.
- a possible implementation manner of the foregoing S203 includes S2031-S2034:
- the actual two-dimensional coordinates of the image feature point P of the target object presented on the N images by the shooting device are denoted by P 1 , P 2 ,..., P N respectively , where the image feature point P is on the i
- the actual two-dimensional coordinates P i on the image are, for example, (u i , v i ).
- S2032 Determine theoretical two-dimensional coordinates of the image feature points in the N images that are related to the three-dimensional coordinates of the image feature points in the world coordinate system.
- the three-dimensional coordinates of the image feature point in the world coordinate system are first predefined, and the three-dimensional coordinates of the image feature point in the world coordinate system are pre-defined (It should be noted that the value of the three-dimensional coordinate is temporarily unknown here) , And the conversion relationship between the three-dimensional coordinates and the two-dimensional coordinates when the image feature points in the three-dimensional coordinate system are respectively projected onto the two-dimensional coordinate system of the above N images, to obtain the theoretical two-dimensional image feature points in each image coordinate.
- the projection angles are different.
- the theoretical two-dimensional coordinates in N images may be different. For example, if the three-dimensional coordinate of the image feature point P of the predefined target object in the world coordinate system is S(x, y, z), then the theoretical two-dimensional coordinate of the image feature point P in the i-th image is u i ', v i ').
- a possible implementation manner of S2032 is: for each image, determine the three-dimensional coordinates of the image feature point in the world coordinate system according to the posture of the shooting device under the camera position corresponding to the image The theoretical two-dimensional coordinates of the image feature point in the image.
- the theoretical two-dimensional coordinates of the image feature points in the image are not only related to the three-dimensional coordinates of the image feature points in the world coordinate system, but also the machine corresponding to the image.
- the position of the camera under the position is related.
- the pose includes a rotation matrix and a displacement matrix.
- the N different camera positions of the camera are expressed as C 1 , C 2 , ..., CN , respectively, and the pose of the camera can be expressed as the image feature point P at the camera position C i with among them, Represents the rotation matrix, Represents the displacement matrix, and G represents the origin of the world coordinate system, which can be determined according to the actual application scenario.
- the actual two-dimensional coordinates in the N images according to the image feature points P are: (u 1 , v 1 ), (u 2 , v 2 ), ..., (u N , v N ) and the image
- the theoretical two-dimensional coordinates of feature points P in N images are: (u 1 ', v 1 '), (u 2 ', v 2 '), ..., (u N ', v N '), and determine the
- the three-dimensional coordinates of the image feature point P in the world coordinate system are P(x, y, z), and it is determined that the theoretical three-dimensional coordinates corresponding to the image feature points in the N images are: (x 1 ', y 1 ', z 1 '), (x 2 ', y 2 ', z 2 '), ..., (x N ', y N ', z N ').
- a possible implementation manner of the foregoing S2033 is: according to the minimum sum of the error between the actual two-dimensional coordinates and the theoretical two-dimensional coordinates of the image feature points in the N images, respectively, determine the The three-dimensional coordinates of the image feature points in the world coordinate system; then according to the three-dimensional coordinates of the image feature points in the world coordinate system, the theoretical three-dimensional coordinates of the image feature points in the N images are determined respectively.
- the theoretical two-dimensional coordinates of the image feature point P on each image can be changed, so that the actual two-dimensional coordinates of the image feature point P on each image and The sum of the errors between the theoretical two-dimensional coordinates changes, so that when the sum of the errors between the actual two-dimensional coordinates and the theoretical two-dimensional coordinates of the image feature point P on each image is the smallest, the corresponding image at this time can be considered
- the three-dimensional coordinates of the feature point P in the world coordinate system are closest to the true three-dimensional coordinates of the image feature point P in the world coordinate system.
- the three-dimensional coordinates can be regarded as the true three-dimensional coordinates of the image feature point P in the world coordinate system.
- the three-dimensional coordinates of the image feature point P in the world coordinate system can be adjusted according to formula 2, to obtain the time when the sum of the errors between the actual two-dimensional coordinates and the theoretical two-dimensional coordinates of the image feature point P on each image is minimum. The specific value of the three-dimensional coordinates.
- the image feature points P can be determined in the N images. Corresponding theoretical three-dimensional coordinates (x 1 ', y 1 ', z 1 '), (x 2 ', y 2 ', z 2 '), ..., (x N ', y N ', z N ').
- a possible implementation manner of determining the theoretical three-dimensional coordinates corresponding to the image feature points in the N images is: for each For an image, determine the theoretical three-dimensional coordinates of the image feature point in the image according to the posture of the camera under the camera position corresponding to the image and the three-dimensional coordinates of the image feature point in the world coordinate system.
- Equation 3 the relationship between the theoretical three-dimensional coordinates of the image feature point P in the i-th image and the theoretical two-dimensional coordinates in the i-th image is shown in Equation 3:
- the relative difference of the image feature points in all N images can be determined.
- a possible implementation manner of S2034 is: for each image, determine the image feature point in the image according to the theoretical three-dimensional coordinates corresponding to the image feature point in the image Corresponding Jacobian matrix; combine the Jacobian matrix corresponding to the image feature points in the N images as column elements in the manner of column vectors to generate a combination matrix; according to the combination matrix, determine the image The relative difference of feature points.
- the Jacobian matrix corresponding to the image feature point in the image is determined according to the theoretical three-dimensional coordinates corresponding to the image feature point in the image.
- the Jacobian matrix is a 2*3 matrix.
- Equation 5 the Jacobian matrix J i of the image feature point P in the i-th image is shown in Equation 5:
- x i ', y i ', z i ' are the coordinate values of the theoretical three-dimensional coordinates corresponding to the image feature point P in the i-th image, Represents the rotation matrix corresponding to the ith image.
- the Jacobian matrix of the image feature point P in each of the N images: J 1 , J 2 ,..., J N can be obtained .
- Equation 6 the Jacobian matrices corresponding to the image feature points in the N images are combined as column elements in the manner of column vectors to generate a combination matrix J, which is shown in Equation 6:
- a possible implementation manner of determining the relative difference of the image feature points P according to the combination matrix is: obtaining a transpose matrix of the combination matrix according to the combination matrix; according to the transposition matrix and the The inverse matrix of the product of the combination matrix determines the relative difference of the image feature points.
- the transpose matrix J T of the combination matrix J the inverse matrix of the product of the transpose matrix and the combination matrix is represented by Q, and the inverse matrix Q is Equation 7:
- the relative difference of the image feature points P is determined.
- the elements on the main diagonal of the inverse matrix are obtained; according to the elements on the main diagonal of the inverse matrix, the relative difference of the image feature points is determined.
- the inverse matrix Q is a square matrix with the same number of rows and columns. Therefore, the elements on the main diagonal of the inverse matrix Q can be obtained. Element to determine the relative difference of the image feature points P.
- a possible implementation of determining the relative difference of the image feature points P is: after summing the elements on the main diagonal of the inverse matrix The value obtained after square rooting is taken as the relative difference of the image feature points.
- Equation 8 the relative difference of the image feature points P can be calculated by Equation 8:
- m is the number of rows or columns of the inverse matrix Q
- DOP represents the relative difference of the image feature points P.
- DOP represents the relative difference of the image feature points P.
- a computer storage medium is also provided in an embodiment of the present invention.
- the computer storage medium stores program instructions, and when the program is executed, it may include a part of the image feature point evaluation method as shown in FIG. 2 and its corresponding embodiments or All steps.
- FIG. 3 is a schematic structural diagram of a movable platform provided by an embodiment of the present invention.
- the movable platform 300 of this embodiment may include: a processor 301 and a shooting device 302. Among them, the processor 301 and the shooting device 302 are connected by a bus.
- the shooting device 302 is used to acquire N images under N different camera positions.
- the processor 301 is configured to acquire N images acquired by the shooting device 302 under N different camera positions, wherein each image includes an image of a target, and N is an integer greater than or equal to 2; Relevant information of the same image feature points of the target object image in the N images; according to the relevant information of the image feature points in the N images, determine the relative difference of the image feature points; The relative difference is evaluated on the image feature points.
- the image feature points are M, and M is an integer greater than or equal to 2.
- the processor 301 evaluates the image feature points according to the relative difference, it is specifically used to:
- the image feature points are evaluated according to the relative differences of the image feature points and the relative differences of other image feature points.
- the processor 301 when the processor 301 evaluates the image feature points according to the relative differences of the image feature points and the relative differences of other image feature points, it is specifically used to:
- the image feature points are evaluated as having high reliability Image feature points;
- K is an integer greater than or equal to 1.
- the processor 301 evaluates the image feature points according to the relative difference, it is specifically used to:
- the image feature points are evaluated as image feature points with high reliability.
- the processor 301 is further used to select image feature points with high reliability for tracking the target.
- the processor 301 determines the relative difference of the image feature points according to the relevant information of the image feature points in the N images, respectively, it is specifically used to:
- the relative difference of the image feature points is determined according to the theoretical three-dimensional coordinates corresponding to the image feature points in the N images, respectively.
- the processor 301 is specifically used to determine the relative difference of the image feature points according to the theoretical three-dimensional coordinates corresponding to the image feature points in the N images, respectively:
- the relative difference of the image feature points is determined.
- the processor 301 determines the relative difference of the image feature points according to the combination matrix, it is specifically used to:
- the relative difference of the image feature points is determined according to the inverse matrix of the product of the transpose matrix and the combination matrix.
- the processor 301 determines the relative difference of the image feature points according to the inverse matrix of the product of the transpose matrix and the combination matrix, it is specifically used to:
- the relative difference of the image feature points is determined according to the elements on the main diagonal of the inverse matrix.
- the processor 301 determines the relative difference of the image feature points according to the elements on the main diagonal of the inverse matrix, it is specifically used to:
- the Jacobian matrix is a 2*3 matrix.
- processor 301 determines the theoretical two-dimensional coordinates of the image feature points related to the three-dimensional coordinates of the image feature points in the world coordinate system in the N images, respectively, specifically to:
- the theoretical two-dimensional coordinates of the image feature point in the image are determined according to the posture of the shooting device 302 under the camera position corresponding to the image and the three-dimensional coordinates of the image feature point in the world coordinate system.
- the processor 301 determines that the image feature points are in the N images according to the actual two-dimensional coordinates and the theoretical two-dimensional coordinates of the image feature points in the N images, respectively.
- the theoretical three-dimensional coordinates corresponding to are specifically used for:
- the theoretical three-dimensional coordinates corresponding to the image feature points in the N images are determined according to the three-dimensional coordinates of the image feature points in the world coordinate system.
- the processor 301 determines the theoretical three-dimensional coordinates corresponding to the image feature points in the N images according to the three-dimensional coordinates of the image feature points in the world coordinate system, specifically Used for:
- the theoretical three-dimensional coordinates of the image feature point in the image are determined according to the posture of the camera 302 under the camera position corresponding to the image and the three-dimensional coordinates of the image feature point in the world coordinate system.
- the pose includes a rotation matrix and a displacement matrix.
- the movable platform of this embodiment may be used to execute the technical solution of the movable platform in the above method embodiments of the present invention, and its implementation principles and technical effects are similar, and will not be repeated here.
- the foregoing program may be stored in a computer-readable storage medium, and when the program is executed, It includes the steps of the above method embodiments; and the foregoing storage media include: read-only memory (Read-Only Memory, ROM), random access memory (Random Access Memory, RAM), magnetic disks or optical discs, etc., which can store program codes Medium.
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
一种图像特征点的评价方法和可移动平台,所述方法包括:获取拍摄装置在N种不同机位下采集获得的N张图像(S201),其中,每张图像包括目标物的图像,所述N为大于等于2的整数;提取目标物的图像的同一图像特征点分别在所述N张图像中的相关信息(S202);根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异(S203);根据所述相对差异,对所述图像特征点进行评价(S204)。因此可以准确地评价出可信度高的图像特征点,可信度高的图像特征点可以用于可移动平台跟踪目标物,从而可以提高跟踪目标物的准确性。
Description
本发明实施例涉及图像处理领域,尤其涉及一种图像特征点的评价方法和可移动平台。
计算机视觉算法中,一般对多帧图像提取目标物的图像特征点,并对多帧图像中提取的图像特征点进行匹配,以计算各图像特征点在世界坐标系下的三维坐标,在获得三维坐标后,应用于无人机领域,可以使得无人机根据三维坐标跟踪该目标物。
目前,在对图像特征点进行匹配时,需要判断该图像特征点的好坏,一般是根据该图像特征点自身的唯一性以及匹配契合程度来判断该图像特征点的好坏。其中,唯一性为在该图像特征点周围没有与其相似的其他图像特征点,匹配契合程度为该图像特征点和其他图像上与其匹配的图像特征点的相似度。
然而,通过上述方式,会造成多帧图像中判断出的好的图像特征点在匹配时实际上并不完全匹配,从而使得计算得到的图像特征点的三维坐标不准确,从而造成无法准确跟踪到目标物。
发明内容
本发明实施例提供一种图像特征点的评价方法和可移动平台,以实现准确地评价出可信度高的图像特征点,使可信度高的图像特征点可以用于可移动平台跟踪目标物,从而可以提高跟踪目标物的准确性。
第一方面,本发明实施例提供一种图像特征点的评价方法,应用于可移动平台,包括:
获取拍摄装置在N种不同机位下采集获得的N张图像,其中,每张图像包括目标物的图像,所述N为大于等于2的整数;
提取目标物的图像的同一图像特征点分别在所述N张图像中的相关信 息;
根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异;
根据所述相对差异,对所述图像特征点进行评价。
第二方面,本发明实施例提供一种可移动平台,包括:处理器和拍摄装置;
所述拍摄装置,用于置在N种不同机位下采集获得N张图像;
所述处理器,用于获取所述拍摄装置在N种不同机位下采集获得的N张图像,其中,每张图像包括目标物的图像,所述N为大于等于2的整数;提取目标物的图像的同一图像特征点分别在所述N张图像中的相关信息;根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异;根据所述相对差异,对所述图像特征点进行评价。
第三方面,本发明实施例提供一种计算机可读存储介质,所述计算机可读存储介质存储有计算机程序,所述计算机程序包含至少一段代码,所述至少一段代码可由计算机执行,以控制所述计算机执行第一方面本发明实施例所述的图像特征点的评价方法。
第四方面,本发明实施例提供一种计算机程序,当所述计算机程序被计算机执行时,用于实现第一方面本发明实施例所述的图像特征点的评价方法。
本发明实施例提供的图像特征点的评价方法和可移动平台,通过获取拍摄装置在N种不同机位下采集获得的N张图像,然后提取目标物的图像的同一图像特征点分别在N张图像中的相关信息,根据图像特征点分别在N张图像中的相关信息,确定图像特征点的相对差异,最后根据相对差异,对图像特征点进行评价。因此可以准确地评价出可信度高的图像特征点,可信度高的图像特征点可以用于可移动平台跟踪目标物,从而可以提高跟踪目标物的准确性。
为了更清楚地说明本发明实施例或现有技术中的技术方案,下面将对实施例或现有技术描述中所需要使用的附图作一简单地介绍,显而易见地,下面描述中的附图是本发明的一些实施例,对于本领域普通技术人员来讲,在 不付出创造性劳动的前提下,还可以根据这些附图获得其他的附图。
图1是根据本发明的实施例的无人飞行系统的示意性架构图;
图2为本发明一实施例提供的图像特征点的评价方法的流程图;
图3为本发明一实施例提供的可移动平台的结构示意图。
为使本发明实施例的目的、技术方案和优点更加清楚,下面将结合本发明实施例中的附图,对本发明实施例中的技术方案进行清楚、完整地描述,显然,所描述的实施例是本发明一部分实施例,而不是全部的实施例。基于本发明中的实施例,本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例,都属于本发明保护的范围。
需要说明的是,当组件被称为“固定于”另一个组件,它可以直接在另一个组件上或者也可以存在居中的组件。当一个组件被认为是“连接”另一个组件,它可以是直接连接到另一个组件或者可能同时存在居中组件。
除非另有定义,本文所使用的所有的技术和科学术语与属于本发明的技术领域的技术人员通常理解的含义相同。本文中在本发明的说明书中所使用的术语只是为了描述具体的实施例的目的,不是旨在于限制本发明。本文所使用的术语“及/或”包括一个或多个相关的所列项目的任意的和所有的组合。
下面结合附图,对本发明的一些实施方式作详细说明。在不冲突的情况下,下述的实施例及实施例中的特征可以相互组合。
本发明的实施例提供了图像特征点的评价方法和可移动平台。该可移动平台例如可以是无人机、无人船、无人汽车、机器人等。其中无人机例如可以是旋翼飞行器(rotorcraft),例如,由多个推动装置通过空气推动的多旋翼飞行器,本发明的实施例并不限于此。
图1是根据本发明的实施例的无人飞行系统的示意性架构图。本实施例以旋翼无人机为例进行说明。
无人飞行系统100可以包括无人机110、显示设备130和控制终端140。其中,无人机110可以包括动力系统150、飞行控制系统160、机架和承载在机架上的云台120。无人机110可以与控制终端140和显示设备130进行无 线通信。
机架可以包括机身和脚架(也称为起落架)。机身可以包括中心架以及与中心架连接的一个或多个机臂,一个或多个机臂呈辐射状从中心架延伸出。脚架与机身连接,用于在无人机110着陆时起支撑作用。
动力系统150可以包括一个或多个电子调速器(简称为电调)151、一个或多个螺旋桨153以及与一个或多个螺旋桨153相对应的一个或多个电机152,其中电机152连接在电子调速器151与螺旋桨153之间,电机152和螺旋桨153设置在无人机110的机臂上;电子调速器151用于接收飞行控制系统160产生的驱动信号,并根据驱动信号提供驱动电流给电机152,以控制电机152的转速。电机152用于驱动螺旋桨旋转,从而为无人机110的飞行提供动力,该动力使得无人机110能够实现一个或多个自由度的运动。在某些实施例中,无人机110可以围绕一个或多个旋转轴旋转。例如,上述旋转轴可以包括横滚轴(Roll)、偏航轴(Yaw)和俯仰轴(pitch)。应理解,电机152可以是直流电机,也可以交流电机。另外,电机152可以是无刷电机,也可以是有刷电机。
飞行控制系统160可以包括飞行控制器161和传感系统162。传感系统162用于测量无人机的姿态信息,即无人机110在空间的位置信息和状态信息,例如,三维位置、三维角度、三维速度、三维加速度和三维角速度等。传感系统162例如可以包括陀螺仪、超声传感器、电子罗盘、惯性测量单元(Inertial Measurement Unit,IMU)、视觉传感器、全球导航卫星系统和气压计等传感器中的至少一种。例如,全球导航卫星系统可以是全球定位系统(Global Positioning System,GPS)。飞行控制器161用于控制无人机110的飞行,例如,可以根据传感系统162测量的姿态信息控制无人机110的飞行。应理解,飞行控制器161可以按照预先编好的程序指令对无人机110进行控制,也可以通过响应来自控制终端140的一个或多个控制指令对无人机110进行控制。
云台120可以包括电机122。云台用于携带拍摄装置123。飞行控制器161可以通过电机122控制云台120的运动。可选地,作为另一实施例,云台120还可以包括控制器,用于通过控制电机122来控制云台120的运动。应理解,云台120可以独立于无人机110,也可以为无人机110的一部分。 应理解,电机122可以是直流电机,也可以是交流电机。另外,电机122可以是无刷电机,也可以是有刷电机。还应理解,云台可以位于无人机的顶部,也可以位于无人机的底部。
拍摄装置123例如可以是照相机或摄像机等用于捕获图像的设备,拍摄装置123可以与飞行控制器通信,并在飞行控制器的控制下进行拍摄。本实施例的拍摄装置123至少包括感光元件,该感光元件例如为互补金属氧化物半导体(Complementary Metal Oxide Semiconductor,CMOS)传感器或电荷耦合元件(Charge-coupled Device,CCD)传感器。可以理解,拍摄装置123也可直接固定于无人机110上,从而云台120可以省略。
显示设备130位于无人飞行系统100的地面端,可以通过无线方式与无人机110进行通信,并且可以用于显示无人机110的姿态信息。另外,还可以在显示设备130上显示拍摄装置拍摄的图像。应理解,显示设备130可以是独立的设备,也可以集成在控制终端140中。
控制终端140位于无人飞行系统100的地面端,可以通过无线方式与无人机110进行通信,用于对无人机110进行远程操纵。
应理解,上述对于无人飞行系统各组成部分的命名仅是出于标识的目的,并不应理解为对本发明的实施例的限制。
因此,无人机110可以识别拍摄装置123拍摄到的图像中的目标物,以对目标物进行跟踪。
图2为本发明一实施例提供的图像特征点的评价方法的流程图,如图2所示,本实施例的方法可以应用于可移动平台,本实施例的方法可以包括:
S201、获取拍摄装置在N种不同机位下采集获得的N张图像。
其中,每张图像包括目标物的图像,所述N为大于等于2的整数。
本实施例中,可移动平台中搭载有拍摄装置,拍摄装置可以用于采集图像,拍摄装置可以在N种不同机位下采集获得N张图像,其中,获得的每张图像中包括目标物的图像,N为大于等于2的整数。例如:拍摄装置在不同的位姿下采集获得上述各图像,不同的位姿可以是拍摄装置处于不同的空间位置,或者,处于不同的旋转角度等,本实施例对此不做限定。
S202、提取目标物的图像的同一图像特征点分别在所述N张图像中的相关信息。
本实施例中,图像特征点为目标物上的特征点。可选的,图像特征点例如为M个,M为大于等于2的整数。例如,针对每个图像特征点,通过拍摄装置在不同机位下拍摄该目标物时,在图像特征点没有被遮挡的情况下,同一个图像特征点(即目标物上同一个特征点)将呈现在每张图像上。可选地,每张图像上的图像特征点的数量可能不同,例如,拍摄装置在不同机位下拍摄该目标物时,由于拍摄装置拍摄角度的变化和/或目标物的位移,导致目标物上的至少一个特征点无法被拍摄装置采集到,从而导致每张图像上的图像特征点(即目标物上的特征点)的数量不同。并且,同一个图像特征点在不同图像上的呈现时的相关信息不同,相关信息例如可以是该图像特征点在不同图像上的位置信息,该位置信息例如可以用图像特征点在图像上的实际二维坐标来表示。其中,实际二维坐标为目标物通过拍摄装置采集,获得目标物的图像后,可以实际测量得到的目标物上的特征点(即图像特征点)在图像上的位置在图像上设定的二维坐标系下的二维坐标。需要说明的是,每张图像上设定二维坐标系的方法相同。
S203、根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异。
S204、根据所述相对差异,对所述图像特征点进行评价。
本实施例中,本实施例中,在获得同一个图像特征点在每张图像上的相关信息之后,根据同一个图像特征点在每张图像上的相关信息,确定该图像特征点在所有N张图像中的相对差异。其中,相对差异例如为该图像特征点在所有N张图像中的实际二维坐标之间的差异度,可以包括方差。然后根据该图像特征点的相对差异,对该图像特征点进行评价。例如:若相对差异小,则该图像特征点为可信度高的图像特征点,若相对差异大,则该图像特征点为可信度低的图像特征点。例如,对于一个图像特征点,当其相对差异为1时,认为该图像特征点为可信度高的图像特征点;当其相对差异为20时,认为该图像特征点为可信度低的图像特征点。然后,本实施例还可以根据可信息度高的图像特征点来跟踪该目标物,以提高跟踪目标物的准确率。
在一些实施例中,由于图像特征点的个数为M个,上述S204的一种可能实现方式为:根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价。本实施例中,在对一个图像特征点进 行评价时,不仅要参考该图像特征点的相对差异,还需要参考其它图像特征点的相对差异。可选的,其它图像特征点可以是其他M-1个图像特征点,也可以是其他M-1个图像特征点中的部分图像特征点,本发明实施例对此不限定。
可选的,根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价的一种可能的实现方式为:若所述图像特征点的相对差异是所述图像特征点以及所述其它图像特征点的相对差异中相对差异最小的前K个图像特征点,则评价所述图像特征点为可信度高的图像特征点。其中,K为大于等于1的整数。另外,K的大小可以根据图像的复杂度、图像的亮度、所处的模式变动,以使选择出可信度高的图像特征点,本实施例对此不做限定。
本实施例中,以其它图像特征点为其它M-1个图像特征点为例进行说明,对M个图像特征点的相对差异进行比较,对这M个图像特征点的相对差异的大小进行排序,然后选取图像特征点的相对差异最小的前K个图像特征点,若该图像特征点包含在图像特征点的相对差异最小的前K个图像特征点中,则说明该图像特征点为可信度高的图像特征点。若该图像特征点不包含在图像特征点的相对差异最小的前K个图像特征点中,则说明该图像特征点为可信度低的图像特征点。
在一些实施例中,上述S204的另一种可能实现方式为:若所述图像特征点的相对差异小于预设差异,则评价所述图像特征点为可信度高的图像特征点。
本实施例中,通过设定预设差异,将图像特征点的相对差异与预设差异进行比较,若该图像特征点的相对差异小于预设差异,则说明该图像特征点为可信度高的图像特征点,若该图像特征点的相对差异大于或等于预设差异,则说明该图像特征点为可信度低的图像特征点。
在一些实施例中,上述S204的另一种可能实现方式为:若所述图像特征点的相对差异小于预设差异,并且所述图像特征点的相对差异是所述图像特征点以及所述其它图像特征点的相对差异中相对差异最小的前K个图像特征点,则评价所述图像特征点为可信度高的图像特征点。
本实施例中,以其它图像特征点为其它M-1个图像特征点为例进行说明, 先将M个图像特征点的相对差异与预设差异进行比较,获取相对差异小于预设差异的至少一个图像特征点。然后,对该至少一个图像特征点的相对差异进行比较,获得图像特征点的相对差异最小的前K个图像特征点,该前K个图像特征点为可信度高的图像特征点。需要说明的是,如果相对差异小于预设差异的图像特征点的数量小于K个,则不需要将相对差异小于预设差异的图像特征点的相对差异进行比较,即可确定相对差异小于预设差异的图像特征点均为可信度高的图像特征点。
本实施例提供的图像特征点的评价方法,通过获取拍摄装置在N种不同机位下采集获得的N张图像,然后提取目标物的图像的同一图像特征点分别在N张图像中的相关信息,根据图像特征点分别在N张图像中的相关信息,确定图像特征点的相对差异,最后根据相对差异,对图像特征点进行评价。因此可以准确地评价出可信度高的图像特征点,可信度高的图像特征点可以用于可移动平台跟踪目标物,从而可以提高跟踪目标物的准确性。
在一些实施例中,上述S203的一种可能实现方式包括S2031-S2034:
S2031、获取所述图像特征点分别在所述N张图像中的实际二维坐标。
本实施例中,目标物的图像特征点P通过拍摄装置呈现在N张图像上的实际二维坐标分别用P
1、P
2、…、P
N表示,其中,图像特征点P在第i张图像上的实际二维坐标P
i例如为(u
i,v
i)。
S2032、确定与所述图像特征点在世界坐标系的三维坐标有关的所述图像特征点分别在所述N张图像中的理论二维坐标。
本实施例中,首先预定义图像特征点在世界坐标系的三维坐标,根据预定义的图像特征点在世界坐标系的三维坐标(需要说明的是,在此处该三维坐标的值暂时未知),以及三维坐标系下的图像特征点在分别投影到上述N张图像中的二维坐标系时三维坐标与二维坐标之间的转换关系,获得图像特征点在每张图像中的理论二维坐标。由于将三维坐标系下的图像特征点投影到不同图像的二维坐标系时,由于不同图像是拍摄装置在不同机位下采集获得的,所以其投影的角度不同,因此,同一图像特征点在N张图像中的理论二维坐标可能不同。例如,预定义目标物的图像特征点P在世界坐标系下的三维坐标为S(x、y、z),则可以获得该图像特征点P在第i张图像中的理论二维坐标为(u
i’,v
i’)。
可选的,S2032的一种可能的实现方式为:针对每张图像,根据该图像对应的机位下的拍摄装置的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论二维坐标。
本实施例中,针对N张图像中的每张图像,图像特征点在该图像中的理论二维坐标不仅与所述图像特征点在世界坐标系的三维坐标有关,还与该图像对应的机位下的拍摄装置的位姿有关。可选的,位姿包括旋转矩阵和位移矩阵。
例如,拍摄装置的N种不同机位分别表示为C
1、C
2、…、C
N,图像特征点P在机位C
i获下拍摄装置的位姿可以表示为
和
其中,
表示旋转矩阵,
表示位移矩阵,G表示世界坐标系的原点,该原点可以根据实际应用场景而定。
图像特征点P在第i张图像中的理论二维坐标可根据如下公式1获得:
S2033、根据所述图像特征点分别在所述N张图像中的实际二维坐标以及理论二维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标。
本实施例中,根据图像特征点P分别在N张图像中的实际二维坐标为:(u
1,v
1)、(u
2,v
2)、…、(u
N,v
N)以及图像特征点P分别在N张图像中的理论二维坐标为:(u
1’,v
1’)、(u
2’,v
2’)、…、(u
N’,v
N’),确定该图像特征点P在世界坐标系中的三维坐标为P(x、y、z),确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标分别为:(x
1’、y
1’、 z
1’)、(x
2’、y
2’、z
2’)、…、(x
N’、y
N’、z
N’)。
在一些实施例中,上述S2033的一种可能的实现方式为:根据所述图像特征点分别在所述N张图像中的实际二维坐标与理论二维坐标的误差之和最小,确定所述图像特征点在世界坐标系下的三维坐标;然后根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标。
具体的,图像特征点P通过拍摄装置投影在每张图像上时,由于存在重投影误差,导致图像特征点P在每张图像上的测量得到的实际二维坐标和计算得到的理论二维坐标之间存在误差。又由于图像特征点P在每张图像上的理论二维坐标根据图像特征点P在世界坐标系下的三维坐标获得。因此,通过调整图像特征点P在世界坐标系下的三维坐标可以改变图像特征点P在每张图像上的理论二维坐标,从而使图像特征点P在每张图像上的实际二维坐标和理论二维坐标之间存在误差之和发生变化,从而当图像特征点P在每张图像上的实际二维坐标和理论二维坐标之间存在误差之和最小时,可以认为此时对应的图像特征点P在世界坐标系下的三维坐标最接近图像特征点P在世界坐标系下真实三维坐标。因此,可将该三维坐标认为是图像特征点P在世界坐标系下的真实三维坐标。其中,可根据公式2调整图像特征点P在世界坐标系下的三维坐标,获得使图像特征点P在每张图像上的实际二维坐标和理论二维坐标之间存在误差之和最小时的三维坐标的具体值。
确定图像特征点P在世界坐标系下的三维坐标的具体值后,根据所述图像特征点在世界坐标系下的三维坐标的具体值,即可确定图像特征点P分别在N张图像中所对应的理论三维坐标(x
1’、y
1’、z
1’)、(x
2’、y
2’、z
2’)、…、(x
N’、y
N’、z
N’)。
可选的,根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标的一种可能的实现方式为:针对每张图像,根据该图像对应的机位下的拍摄装置的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论三维坐标。
其中,已获得图像特征点P在世界坐标系下的三维坐标(x、y、z)的具 体值,其在第i张图像中的理论三维坐标为(x
i’、y
i’、z
i’),图像特征点P在第i张图像中的理论二维坐标为(u
i’,v
i’)。另外,图像特征点P在第i张图像中的理论三维坐标与其在第i张图像中的理论二维坐标之间的关系如公式3所示:
S2034、根据所述图像特征点分别在所述N张图像中所对应的理论三维坐标,确定所述图像特征点的相对差异。
在获得图像特征点分别在所述N张图像中所对应的理论三维坐标之后,根据这些理论三维坐标,可确定图像特征点在所有N张图像中的相对差异。
在一些实施例中,S2034的一种可能的实现方式为:针对每张图像,根据所述图像特征点在该图像中所对应的理论三维坐标,确定所述图像特征点在所述图像中所对应的雅克比矩阵;将所述图像特征点分别在所述N张图像中所对应的雅克比矩阵作为列元素按照列向量的方式组合,生成组合矩阵;根据所述组合矩阵,确定所述图像特征点的相对差异。
本实施例中,针对N张图像中的每张图像,根据所述图像特征点在该图像中所对应的理论三维坐标,确定所述图像特征点在所述图像中所对应的雅克比矩阵。可选地,雅克比矩阵为2*3的矩阵。
具体的,图像特征点P在第i张图像中的雅克比矩阵J
i如公式5所示:
其中,x
i’、y
i’、z
i’分别为图像特征点P在第i张图像中所对应的理论三维坐标的坐标值,
表示第i张图像对应的旋转矩阵。通过上述方式,可以获得图像特征点P在N张图像中每张图像中的雅克比矩阵:J
1、J
2、…、J
N。
再将所述图像特征点分别在所述N张图像中所对应的雅克比矩阵作为列元素按照列向量的方式组合,生成组合矩阵J,组合矩阵J如公式6所示:
然后根据组合矩阵,确定图像特征点P的相对差异。可选地,根据组合矩阵,确定图像特征点P的相对差异的一种可能的实现方式为:根据所述组合矩阵,获得所述组合矩阵的转置矩阵;根据所述转置矩阵与所述组合矩阵的乘积的逆矩阵,确定所述图像特征点的相对差异。
本实施例中,组合矩阵J的转置矩阵J
T,则转置矩阵与组合矩阵的乘积的逆矩阵用Q表示,则逆矩阵Q为公式7:
Q=(J
TJ)
-1 公式7
然后根据上述逆矩阵Q,确定图像特征点P的相对差异。例如:在获得逆矩阵Q之后,获取所述逆矩阵的主对角线上的元素;根据所述逆矩阵的主对角线上的元素,确定所述图像特征点的相对差异。本实施例,根据公式7可知,逆矩阵Q为行数和列数相等的方阵,因此,可获取逆矩阵Q的主对角线上的元素,根据逆矩阵Q的主对角线上的元素,确定图像特征点P的相对差异。
可选的,根据逆矩阵Q的主对角线上的元素,确定图像特征点P的相对差异的一种可能的实现方式为:将所述逆矩阵的主对角线上的元素求和后开平方后获得的值作为所述图像特征点的相对差异。
本实施例中,逆矩阵Q的主对角线上的元素Q
ii,其中,j为元素Q
ii所在的行数或列数。因此,图像特征点P的相对差异可由公式8计算获得:
其中,m为逆矩阵Q的行数或列数,DOP表示图像特征点P的相对差异。
可选的,确定图像特征点P在世界坐标系下的三维坐标的具体值后,根据所述图像特征点在世界坐标系下的三维坐标的具体值,确定图像特征点P分别在N张图像中所对应的理论二维坐标:(u
1’,v
1’)、(u
2’,v
2’)、…、(u
N’,v
N’)。然后计算图像特征点P在每张图像上的实际二维坐标和理论二维坐标之间的差异,并根据公式9获得差异之和(即图像特征点P的相对差异):
其中,DOP表示图像特征点P的相对差异。
本发明实施例中还提供了一种计算机存储介质,该计算机存储介质中存储有程序指令,所述程序执行时可包括如图2及其对应实施例中的图像特征点的评价方法的部分或全部步骤。
图3为本发明一实施例提供的可移动平台的结构示意图,如图3所示,本实施例的可移动平台300可以包括:处理器301和拍摄装置302。其中,处理器301和拍摄装置302通过总线连接。
所述拍摄装置302,用于在N种不同机位下采集获得N张图像。
所述处理器301,用于获取所述拍摄装置302在N种不同机位下采集获得的N张图像,其中,每张图像包括目标物的图像,所述N为大于等于2的整数;提取目标物的图像的同一图像特征点分别在所述N张图像中的相关信息;根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异;根据所述相对差异,对所述图像特征点进行评价。
在一些实施例中,所述图像特征点为M个,M为大于等于2的整数,所述处理器301在根据所述相对差异,对所述图像特征点进行评价时,具体用于:
根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价。
在一些实施例中,所述处理器301在根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价时,具体用于:
若所述图像特征点的相对差异是所述图像特征点以及所述其它图像特征点的相对差异中相对差异最小的前K个图像特征点,则评价所述图像特征点为可信度高的图像特征点;
其中,K为大于等于1的整数。
所述处理器301在根据所述相对差异,对所述图像特征点进行评价时,具体用于:
若所述图像特征点的相对差异小于预设差异,则评价所述图像特征点为可信度高的图像特征点。
在一些实施例中,所述处理器301,还用于选择可信度高的图像特征点用于跟踪所述目标物。
在一些实施例中,所述处理器301在根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异时,具体用于:
获取所述图像特征点分别在所述N张图像中的实际二维坐标;
确定与所述图像特征点在世界坐标系的三维坐标有关的所述图像特征点分别在所述N张图像中的理论二维坐标;
根据所述图像特征点分别在所述N张图像中的实际二维坐标以及理论二维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标;
根据所述图像特征点分别在所述N张图像中所对应的理论三维坐标,确定所述图像特征点的相对差异。
在一些实施例中,所述处理器301在根据所述图像特征点分别在所述N张图像中所对应的理论三维坐标,确定所述图像特征点的相对差异时,具体用于:
针对每张图像,根据所述图像特征点在该图像中所对应的理论三维坐标,确定所述图像特征点在所述图像中所对应的雅克比矩阵;
将所述图像特征点分别在所述N张图像中所对应的雅克比矩阵作为列元素按照列向量的方式组合,生成组合矩阵;
根据所述组合矩阵,确定所述图像特征点的相对差异。
在一些实施例中,所述处理器301在根据所述组合矩阵,确定所述图像特征点的相对差异时,具体用于:
根据所述组合矩阵,获得所述组合矩阵的转置矩阵;
根据所述转置矩阵与所述组合矩阵的乘积的逆矩阵,确定所述图像特征点的相对差异。
在一些实施例中,所述处理器301在根据所述转置矩阵与所述组合矩阵 的乘积的逆矩阵,确定所述图像特征点的相对差异时,具体用于:
获取所述逆矩阵的主对角线上的元素;
根据所述逆矩阵的主对角线上的元素,确定所述图像特征点的相对差异。
在一些实施例中,所述处理器301根据所述逆矩阵的主对角线上的元素,确定所述图像特征点的相对差异时,具体用于:
将所述逆矩阵的主对角线上的元素求和后开平方后获得的值作为所述图像特征点的相对差异。
在一些实施例中,所述雅克比矩阵为2*3的矩阵。
在一些实施例中,所述处理器301在确定与所述图像特征点在世界坐标系的三维坐标有关的所述图像特征点分别在所述N张图像中的理论二维坐标时,具体用于:
针对每张图像,根据该图像对应的机位下的拍摄装置302的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论二维坐标。
在一些实施例中,所述处理器301在根据所述图像特征点分别在所述N张图像中的实际二维坐标以及理论二维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标时,具体用于:
根据所述图像特征点分别在所述N张图像中的实际二维坐标与理论二维坐标的误差之和最小,确定所述图像特征点在世界坐标系下的三维坐标;
根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标。
在一些实施例中,所述处理器301在根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标时,具体用于:
针对每张图像,根据该图像对应的机位下的拍摄装置302的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论三维坐标。
在一些实施例中,所述位姿包括旋转矩阵和位移矩阵。
本实施例的可移动平台,可以用于执行本发明上述各方法实施例中可移动平台的技术方案,其实现原理和技术效果类似,此处不再赘述。
本领域普通技术人员可以理解:实现上述方法实施例的全部或部分步骤可以通过程序指令相关的硬件来完成,前述的程序可以存储于一计算机可读取存储介质中,该程序在执行时,执行包括上述方法实施例的步骤;而前述的存储介质包括:只读内存(Read-Only Memory,ROM)、随机存取存储器(Random Access Memory,RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
最后应说明的是:以上各实施例仅用以说明本发明的技术方案,而非对其限制;尽管参照前述各实施例对本发明进行了详细的说明,本领域的普通技术人员应当理解:其依然可以对前述各实施例所记载的技术方案进行修改,或者对其中部分或者全部技术特征进行等同替换;而这些修改或者替换,并不使相应技术方案的本质脱离本发明各实施例技术方案的范围。
Claims (31)
- 一种图像特征点的评价方法,其特征在于,包括:获取拍摄装置在N种不同机位下采集获得的N张图像,其中,每张图像包括目标物的图像,所述N为大于等于2的整数;提取目标物的图像的同一图像特征点分别在所述N张图像中的相关信息;根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异;根据所述相对差异,对所述图像特征点进行评价。
- 根据权利要求1所述的方法,其特征在于,所述图像特征点为M个,M为大于等于2的整数,所述根据所述相对差异,对所述图像特征点进行评价,包括:根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价。
- 根据权利要求2所述的方法,其特征在于,所述根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价,包括:若所述图像特征点的相对差异是所述图像特征点以及所述其它图像特征点的相对差异中相对差异最小的前K个图像特征点,则评价所述图像特征点为可信度高的图像特征点;其中,K为大于等于1的整数。
- 根据权利要求1所述的方法,其特征在于,所述根据所述相对差异,对所述图像特征点进行评价,包括:若所述图像特征点的相对差异小于预设差异,则评价所述图像特征点为可信度高的图像特征点。
- 根据权利要求3或4所述的方法,其特征在于,还包括:选择可信度高的图像特征点用于跟踪所述目标物。
- 根据权利要求1-5任一项所述的方法,其特征在于,所述根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异,包括:获取所述图像特征点分别在所述N张图像中的实际二维坐标;确定与所述图像特征点在世界坐标系的三维坐标有关的所述图像特征点分别在所述N张图像中的理论二维坐标;根据所述图像特征点分别在所述N张图像中的实际二维坐标以及理论二维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标;根据所述图像特征点分别在所述N张图像中所对应的理论三维坐标,确定所述图像特征点的相对差异。
- 根据权利要求6所述的方法,其特征在于,所述根据所述图像特征点分别在所述N张图像中所对应的理论三维坐标,确定所述图像特征点的相对差异,包括:针对每张图像,根据所述图像特征点在该图像中所对应的理论三维坐标,确定所述图像特征点在所述图像中所对应的雅克比矩阵;将所述图像特征点分别在所述N张图像中所对应的雅克比矩阵作为列元素按照列向量的方式组合,生成组合矩阵;根据所述组合矩阵,确定所述图像特征点的相对差异。
- 根据权利要求7所述的方法,其特征在于,所述根据所述组合矩阵,确定所述图像特征点的相对差异,包括:根据所述组合矩阵,获得所述组合矩阵的转置矩阵;根据所述转置矩阵与所述组合矩阵的乘积的逆矩阵,确定所述图像特征点的相对差异。
- 根据权利要求8所述的方法,其特征在于,所述根据所述转置矩阵与所述组合矩阵的乘积的逆矩阵,确定所述图像特征点的相对差异,包括:获取所述逆矩阵的主对角线上的元素;根据所述逆矩阵的主对角线上的元素,确定所述图像特征点的相对差异。
- 根据权利要求9所述的方法,其特征在于,所述根据所述逆矩阵的主对角线上的元素,确定所述图像特征点的相对差异,包括:将所述逆矩阵的主对角线上的元素求和后开平方后获得的值作为所述图像特征点的相对差异。
- 根据权利要求7-10任一项所述的方法,其特征在于,所述雅克比矩阵为2*3的矩阵。
- 根据权利要求6所述的方法,其特征在于,所述确定与所述图像特征点在世界坐标系的三维坐标有关的所述图像特征点分别在所述N张图像中的理论二维坐标,包括:针对每张图像,根据该图像对应的机位下的拍摄装置的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论二维坐标。
- 根据权利要求6所述的方法,其特征在于,所述根据所述图像特征点分别在所述N张图像中的实际二维坐标以及理论二维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标,包括:根据所述图像特征点分别在所述N张图像中的实际二维坐标与理论二维坐标的误差之和最小,确定所述图像特征点在世界坐标系下的三维坐标;根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标。
- 根据权利要求13所述的方法,其特征在于,所述根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标,包括:针对每张图像,根据该图像对应的机位下的拍摄装置的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论三维坐标。
- 根据权利要求12或14所述的方法,其特征在于,所述位姿包括旋转矩阵和位移矩阵。
- 一种可移动平台,其特征在于,包括:处理器和拍摄装置;所述拍摄装置,用于置在N种不同机位下采集获得N张图像;所述处理器,用于获取所述拍摄装置在N种不同机位下采集获得的N张图像,其中,每张图像包括目标物的图像,所述N为大于等于2的整数;提取目标物的图像的同一图像特征点分别在所述N张图像中的相关信息;根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异;根据所述相对差异,对所述图像特征点进行评价。
- 根据权利要求16所述的可移动平台,其特征在于,所述图像特征点为M个,M为大于等于2的整数,所述处理器在根据所述相对差异,对所述 图像特征点进行评价时,具体用于:根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价。
- 根据权利要求17所述的可移动平台,其特征在于,所述处理器在根据所述图像特征点的相对差异,以及其它图像特征点的相对差异,对所述图像特征点进行评价时,具体用于:若所述图像特征点的相对差异是所述图像特征点以及所述其它图像特征点的相对差异中相对差异最小的前K个图像特征点,则评价所述图像特征点为可信度高的图像特征点;其中,K为大于等于1的整数。
- 根据权利要求16所述的可移动平台,其特征在于,所述处理器在根据所述相对差异,对所述图像特征点进行评价时,具体用于:若所述图像特征点的相对差异小于预设差异,则评价所述图像特征点为可信度高的图像特征点。
- 根据权利要求18或19所述的可移动平台,其特征在于,所述处理器还用于,选择可信度高的图像特征点用于跟踪所述目标物。
- 根据权利要求16-20任一项所述的可移动平台,其特征在于,所述处理器在根据所述图像特征点分别在所述N张图像中的相关信息,确定所述图像特征点的相对差异时,具体用于:获取所述图像特征点分别在所述N张图像中的实际二维坐标;确定与所述图像特征点在世界坐标系的三维坐标有关的所述图像特征点分别在所述N张图像中的理论二维坐标;根据所述图像特征点分别在所述N张图像中的实际二维坐标以及理论二维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标;根据所述图像特征点分别在所述N张图像中所对应的理论三维坐标,确定所述图像特征点的相对差异。
- 根据权利要求21所述的可移动平台,其特征在于,所述处理器在根据所述图像特征点分别在所述N张图像中所对应的理论三维坐标,确定所述图像特征点的相对差异时,具体用于:针对每张图像,根据所述图像特征点在该图像中所对应的理论三维坐标, 确定所述图像特征点在所述图像中所对应的雅克比矩阵;将所述图像特征点分别在所述N张图像中所对应的雅克比矩阵作为列元素按照列向量的方式组合,生成组合矩阵;根据所述组合矩阵,确定所述图像特征点的相对差异。
- 根据权利要求22所述的可移动平台,其特征在于,所述处理器在根据所述组合矩阵,确定所述图像特征点的相对差异时,具体用于:根据所述组合矩阵,获得所述组合矩阵的转置矩阵;根据所述转置矩阵与所述组合矩阵的乘积的逆矩阵,确定所述图像特征点的相对差异。
- 根据权利要求23所述的可移动平台,其特征在于,所述处理器在根据所述转置矩阵与所述组合矩阵的乘积的逆矩阵,确定所述图像特征点的相对差异时,具体用于:获取所述逆矩阵的主对角线上的元素;根据所述逆矩阵的主对角线上的元素,确定所述图像特征点的相对差异。
- 根据权利要求24所述的可移动平台,其特征在于,所述处理器根据所述逆矩阵的主对角线上的元素,确定所述图像特征点的相对差异时,具体用于:将所述逆矩阵的主对角线上的元素求和后开平方后获得的值作为所述图像特征点的相对差异。
- 根据权利要求22-25任一项所述的可移动平台,其特征在于,所述雅克比矩阵为2*3的矩阵。
- 根据权利要求21所述的可移动平台,其特征在于,所述处理器在确定与所述图像特征点在世界坐标系的三维坐标有关的所述图像特征点分别在所述N张图像中的理论二维坐标时,具体用于:针对每张图像,根据该图像对应的机位下的拍摄装置的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论二维坐标。
- 根据权利要求21所述的可移动平台,其特征在于,所述处理器在根据所述图像特征点分别在所述N张图像中的实际二维坐标以及理论二维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标时, 具体用于:根据所述图像特征点分别在所述N张图像中的实际二维坐标与理论二维坐标的误差之和最小,确定所述图像特征点在世界坐标系下的三维坐标;根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标。
- 根据权利要求28所述的可移动平台,其特征在于,所述处理器在根据所述图像特征点在世界坐标系下的三维坐标,确定所述图像特征点分别在所述N张图像中所对应的理论三维坐标时,具体用于:针对每张图像,根据该图像对应的机位下的拍摄装置的位姿以及所述图像特征点在世界坐标系的三维坐标,确定所述图像特征点在该图像中的理论三维坐标。
- 根据权利要求27或29所述的可移动平台,其特征在于,所述位姿包括旋转矩阵和位移矩阵。
- 一种计算机可读存储介质,其特征在于,所述计算机可读存储介质存储有计算机程序,所述计算机程序包含至少一段代码,所述至少一段代码可由计算机执行,以控制所述计算机执行根据权利要求1-15任一项所述的图像特征点的评价方法。
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2018/118778 WO2020107480A1 (zh) | 2018-11-30 | 2018-11-30 | 图像特征点的评价方法和可移动平台 |
| CN201880073410.5A CN111433815A (zh) | 2018-11-30 | 2018-11-30 | 图像特征点的评价方法和可移动平台 |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| PCT/CN2018/118778 WO2020107480A1 (zh) | 2018-11-30 | 2018-11-30 | 图像特征点的评价方法和可移动平台 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| WO2020107480A1 true WO2020107480A1 (zh) | 2020-06-04 |
Family
ID=70854728
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| PCT/CN2018/118778 Ceased WO2020107480A1 (zh) | 2018-11-30 | 2018-11-30 | 图像特征点的评价方法和可移动平台 |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN111433815A (zh) |
| WO (1) | WO2020107480A1 (zh) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105046749A (zh) * | 2015-09-10 | 2015-11-11 | 深圳市神州龙资讯服务有限公司 | 三视角航片自动生成三维模型的方法 |
| US20160178936A1 (en) * | 2014-12-23 | 2016-06-23 | Multimedia Image Solution Limited | Method of Virtually Trying on Eyeglasses |
| US20170180627A1 (en) * | 2015-12-16 | 2017-06-22 | SK Hynix Inc. | Auto-focus system for a digital imaging device and method |
| CN108304758A (zh) * | 2017-06-21 | 2018-07-20 | 腾讯科技(深圳)有限公司 | 人脸特征点跟踪方法及装置 |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104424651A (zh) * | 2013-08-26 | 2015-03-18 | 株式会社理光 | 跟踪对象的方法和系统 |
| CN107563959B (zh) * | 2017-08-30 | 2021-04-30 | 北京林业大学 | 全景图生成方法及装置 |
-
2018
- 2018-11-30 WO PCT/CN2018/118778 patent/WO2020107480A1/zh not_active Ceased
- 2018-11-30 CN CN201880073410.5A patent/CN111433815A/zh active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160178936A1 (en) * | 2014-12-23 | 2016-06-23 | Multimedia Image Solution Limited | Method of Virtually Trying on Eyeglasses |
| CN105046749A (zh) * | 2015-09-10 | 2015-11-11 | 深圳市神州龙资讯服务有限公司 | 三视角航片自动生成三维模型的方法 |
| US20170180627A1 (en) * | 2015-12-16 | 2017-06-22 | SK Hynix Inc. | Auto-focus system for a digital imaging device and method |
| CN108304758A (zh) * | 2017-06-21 | 2018-07-20 | 腾讯科技(深圳)有限公司 | 人脸特征点跟踪方法及装置 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111433815A (zh) | 2020-07-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109074168B (zh) | 无人机的控制方法、设备和无人机 | |
| US11798172B2 (en) | Maximum temperature point tracking method, device and unmanned aerial vehicle | |
| WO2020113423A1 (zh) | 目标场景三维重建方法、系统及无人机 | |
| CN108235815B (zh) | 摄像控制装置、摄像装置、摄像系统、移动体、摄像控制方法及介质 | |
| CN108762324A (zh) | 云台电机角度和角速度估算方法、装置、云台及飞行器 | |
| CN106525074A (zh) | 一种云台漂移的补偿方法、装置、云台和无人机 | |
| WO2021081774A1 (zh) | 一种参数优化方法、装置及控制设备、飞行器 | |
| CN111316185A (zh) | 可移动平台的巡检控制方法和可移动平台 | |
| WO2019189381A1 (ja) | 移動体、制御装置、および制御プログラム | |
| WO2021217371A1 (zh) | 可移动平台的控制方法和装置 | |
| WO2019144300A1 (zh) | 目标检测方法、装置和可移动平台 | |
| US20210229810A1 (en) | Information processing device, flight control method, and flight control system | |
| WO2019104571A1 (zh) | 图像处理方法和设备 | |
| WO2019183789A1 (zh) | 无人机的控制方法、装置和无人机 | |
| CN111247389A (zh) | 关于拍摄设备的数据处理方法、装置及图像处理设备 | |
| CN110800023A (zh) | 图像处理方法和设备、摄像装置以及无人机 | |
| WO2021043214A1 (zh) | 一种标定方法、装置及飞行器 | |
| US20210256732A1 (en) | Image processing method and unmanned aerial vehicle | |
| WO2020048365A1 (zh) | 飞行器的飞行控制方法、装置、终端设备及飞行控制系统 | |
| WO2021056503A1 (zh) | 可移动平台的定位方法、装置、可移动平台及存储介质 | |
| WO2022141123A1 (zh) | 可移动平台及其控制方法、装置、终端设备和存储介质 | |
| WO2020019260A1 (zh) | 磁传感器校准方法、控制终端以及可移动平台 | |
| CN114096931A (zh) | 可移动平台的控制方法和设备 | |
| CN110869787A (zh) | 磁传感器校准方法以及可移动平台 | |
| WO2020113417A1 (zh) | 目标场景三维重建方法、系统及无人机 |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 18941548 Country of ref document: EP Kind code of ref document: A1 |
|
| NENP | Non-entry into the national phase |
Ref country code: DE |
|
| 122 | Ep: pct application non-entry in european phase |
Ref document number: 18941548 Country of ref document: EP Kind code of ref document: A1 |







