CN-121977508-A - Method for measuring non-cooperative targets in field scene

CN121977508ACN 121977508 ACN121977508 ACN 121977508ACN-121977508-A

Abstract

The invention provides a field scene non-cooperative target measurement method which comprises the steps of firstly, an unmanned aerial vehicle hovering to shoot multi-frame field scene images and obtaining self-positioning geographic coordinates of pixels corresponding to cooperative targets at corresponding moments in each frame of images to establish a control point set between the coordinates, secondly, fitting the control point set by using a quadric surface function to obtain a mapping relation of mutual mapping between coordinates of any pixels in the images based on fitting of the control points, and thirdly, calculating to obtain geographic coordinates of the non-cooperative targets according to the mapping relation and realizing positioning measurement of the non-cooperative targets of the field scene according to the pixel coordinates of the non-cooperative targets in aerial images. According to the invention, the real-time position of the non-cooperative targets such as the unmanned aerial vehicle is used as the control point of the unmanned aerial vehicle aerial image in the field scene, so that the positioning detection of the non-cooperative targets on the flat ground in the image in the field scene is realized.

Inventors

LUO HUAJUN
DU ZEXIN
LUO SHUQI
DING QINGWEI
ZHAO YUN
HU JIE
WEN BIN
LUO GANG
ZHU XINXIN
CHANG LIANG
MA BIN
SHAO XIANG

Assignees

中国人民解放军63618部队

Dates

Publication Date: 20260505
Application Date: 20251229

Claims (6)

1. The field scene non-cooperative target measurement method is characterized by comprising the following steps of: firstly, an unmanned aerial vehicle hovers to shoot a field scene image containing a cooperative target and a non-cooperative target which move in a camera field of view, and establishes a control point set of the cooperative target in the field scene image based on pixel coordinates of the cooperative target in a plurality of frames of field scene images and self-positioning geographic coordinates of the cooperative target in a corresponding image; fitting the control point set obtained in the first step by using a quadric function to obtain a mapping relation between any pixel coordinate and geographic coordinate in the aerial image; And thirdly, calculating the geographic coordinates of the non-cooperative targets according to the pixel coordinates of the non-cooperative targets in the aerial image by utilizing the mapping relation obtained in the second step, and realizing positioning measurement of the non-cooperative targets in the field scene.
2. The method according to claim 1, wherein in the first step, the step of creating the set of control points of the cooperative target in the wild scene image based on the pixel coordinates of the cooperative target in the multiple frames of the wild scene image and the geographic coordinates of the cooperative target in the corresponding frames of the image includes: when the unmanned aerial vehicle hovers and shoots an area containing a cooperative target and a non-cooperative target, acquiring pixel coordinates of the cooperative target in the aerial image in a maneuvering state of the cooperative target and accurate self-positioning geographic coordinates of the cooperative target at the corresponding moment of the image, and taking the accurate self-positioning geographic coordinates as a control point in the field scene image; And continuously acquiring pixel coordinates and high-precision geographic coordinates of the cooperative target at different positions in the image from a plurality of frames of field scene images, and establishing a control point set between the pixel coordinates of the cooperative target in the field scene images and the geographic coordinates corresponding to the pixels.
3. The method for measuring non-cooperative targets in a field scene according to claim 1, wherein in the first step, a set of control points from pixel coordinates to geographic coordinates of the cooperative targets in the field scene image is established, specifically comprising: Shooting field scene images of a cooperative target in a target area under a continuous maneuvering state by adopting an aerial camera in an unmanned aerial vehicle hovering state, thereby obtaining a plurality of field scene images which are ordered in time, and obtaining geographic coordinates of the cooperative target in the field scene images corresponding to the same moment, wherein the selection criteria of the plurality of field scene images are that if the cooperative target existing in the field scene images at the current moment is detected and pixels of the cooperative target and the field scene images of the last control point move on the field scene images, and the moving pixels are greater than a preset threshold value, the pixel coordinates of the cooperative target are extracted from the field scene images at the current moment, and the geographic coordinates of the cooperative target in the field scene images are obtained at the same time and are used as a control point, otherwise, the cooperative target is continuously detected from the next frame of images; and establishing a control point set between pixel coordinates and geographic coordinates of the cooperative target at a plurality of corresponding moments one by utilizing a plurality of frames of field scene images, so that the second step is convenient to fit the control point set by utilizing a quadric function.
4. The method for measuring a non-cooperative target in a field scene according to claim 1, wherein the second step further comprises: And (3) fitting the control point set obtained in the step (I) by using a quadric function, and solving fitting parameters by using a least square method so as to obtain the mapping relation from the pixel coordinates of any pixel in the aerial region image to the geographic coordinates.
5. The method for measuring non-cooperative targets in a field scene according to claim 1 or 4, wherein in the second step, an accurate mapping relationship between any pixel coordinate and geographic coordinate is constructed in a field scene image captured by hovering, and the method specifically comprises: (1) The fitting function to the set of control points between pixel coordinates and geographic coordinates is expressed as: In the formula, Longitude values representing the geographic coordinates, Representing pixel coordinates The value of the vector is used to determine, Representing pixel coordinates The value of the vector is used to determine, A latitude value representing a geographic coordinate, Fitting coefficients to be solved are respectively; (2) Longitude and latitude in the geographic coordinates all need a plurality of groups of non-collinear characteristic points, and the best fitting parameters are solved by using a least square method; (3) Based on the control point set and fitting function, adopting least square method to solve fitting parameter, concretely, for longitude, based on the control point set from pixel coordinate to geographic coordinate, establishing observation vector Sum matrix Solving the parameter vector of the fitting function Obtaining fitting function, calculating longitude data corresponding to any pixel point, assuming n observation points exist, and corresponding pixel coordinates of longitude L The following relationship is satisfied: In the formula, Representing observed data, parameter vectors , The observation vector of the group of non-collinear feature points is By least squares, the best fit parameters are expressed as: Wherein the matrix ; For latitudes, the observation vector is expressed as The best fit parameters are expressed as 。 (4) And (3) obtaining fitting parameters of the fitting functions of the longitude and the latitude based on the step (3), and further obtaining the mapping relation between each pixel coordinate of the real-time frame and the corresponding geographic coordinate in the aerial field scene image.
6. The method for measuring non-cooperative targets in a field scene according to claim 1, wherein said step three comprises: shooting an image of a non-cooperative target in a target area under a hovering state of the unmanned aerial vehicle, and acquiring real-time pixel coordinates of the non-cooperative target in the target area; and obtaining geographic coordinates of the non-cooperative targets corresponding to the pixel coordinates in the third step in real time according to the mapping relation of the second step.

Description

Method for measuring non-cooperative targets in field scene Technical Field The invention belongs to the technical field of aerial photography target measurement, and particularly relates to a field scene non-cooperative target measurement method. Background Unmanned aerial vehicle aerial photography is widely applied to the fields of national surveillance, environmental protection, military reconnaissance, target search and the like, and target positioning in images by using aerial images of unmanned aerial vehicles is a hotspot in the fields of unmanned aerial vehicle aerial photography and target positioning. Currently, a target positioning method based on aerial images of an unmanned aerial vehicle mainly comprises a binocular vision positioning method, a double-machine intersection measuring method, a coordinate transformation method depending on attitude information of the unmanned aerial vehicle, a target positioning method based on scene matching and the like. The binocular vision positioning method is suitable for a close range target, the camera internal and external parameters (focal length, baseline distance and distortion) of the binocular vision positioning method need to be calibrated with high precision, errors are large when the binocular vision positioning method is used in a long distance, the accurate unmanned aerial vehicle position, attitude and cradle head pointing information and other information are needed by the binocular vision measurement method and the coordinate transformation method depending on unmanned aerial vehicle attitude information, the binocular vision positioning method is only suitable for an unmanned aerial vehicle carrying a high-precision cradle head, the cost requirement is high, the target positioning method based on scene matching is sensitive to environment changes, and the binocular vision positioning method is not suitable for repeated textures and weak feature scenes such as gobi, desert, ocean and the like. Disclosure of Invention The invention aims to solve the defects of calibration dependence, scene dependence and the like existing in non-cooperative target positioning measurement based on aerial images, and provides a field scene non-cooperative target measurement method, wherein the real-time position of a cooperative target such as an unmanned vehicle is used as a control point of an aerial image of the unmanned aerial vehicle in a field scene, and the positioning and target detection of the non-cooperative target on a flat ground in the aerial image in the field scene are realized through calculation. In order to achieve the above purpose, the technical solution provided by the present invention is: A field scene non-cooperative target measurement method comprises the following steps: firstly, an unmanned aerial vehicle hovers to shoot a field scene image containing a cooperative target and a non-cooperative target which move in a camera field of view, and establishes a control point set of the cooperative target in the field scene image based on pixel coordinates of the cooperative target in a plurality of frames of field scene images and self-positioning geographic coordinates of the cooperative target in a corresponding image; fitting the control point set obtained in the first step by using a quadric function to obtain a mapping relation between any pixel coordinate and geographic coordinate in the aerial image; and thirdly, calculating to obtain geographic coordinates of the non-cooperative targets according to the pixel coordinates of the non-cooperative targets in the aerial image and the mapping relation obtained in the second step, so as to realize positioning measurement of the non-cooperative targets in the field scene. As a further limitation of the present invention, in the first step, a control point set of the cooperative target in the wild scene image is established based on the pixel coordinates of the cooperative target in the multiple frames of wild scene images and the geographic coordinates of the cooperative target in the corresponding frame images, and the method further includes: when the unmanned aerial vehicle hovers and shoots an area containing a cooperative target and a non-cooperative target, acquiring pixel coordinates of the cooperative target in the aerial image in a maneuvering state of the cooperative target and accurate self-positioning geographic coordinates of the cooperative target at the corresponding moment of the image, and taking the accurate self-positioning geographic coordinates as a control point in the field scene image; And continuously acquiring pixel coordinates and high-precision geographic coordinates of the cooperative target at different positions in the image from a plurality of frames of field scene images, and establishing a control point set between the pixel coordinates of the cooperative target in the field scene images and the geographic coordinates corresponding to the pixels. As a fur