光学成像载荷空间分辨率的球面靶标直接检测法

Direct Detection Method for Spatial Resolution of Optical Imaging Loads Based on Spherical Target

空间分辨率是衡量遥感载荷光学成像性能的重要指标。传统二维扇形靶标影像难以直接用于载荷空间分辨率检测,精度受限于载荷构像模型以及影像校正处理方法。本文结合圆球球面数学模型,设计了扇形与网格混合的球面靶标,提出了一种直接在球面靶标影像上进行光学成像载荷空间分辨率检测方法。该方法针对球面同心圆移动椭圆成像以及偏心率局部一致特点,沿球面同心圆圆心直线方向进行多尺度椭圆拟合,并通过多尺度拟合椭圆长半轴的相对比例关系来精确计算遥感载荷空间分辨率,达到直接基于球面靶标中心投影影像进行载荷空间分辨率检测的效果。采用仿真球面靶标进行检测方法准确性验证,实验结果表明,本文方法具有优于传统检测方法的检测精度。进一步采用中国(嵩山)卫星遥感定标场中布设的实体球面靶标,开展了光学成像载荷空间分辨率检测实验,验证了本文方法的空间分辨率直接检测能力和检测结果有效性。

Objective In the context of intelligent surveying and mapping,the quality of remote sensing imagery directly determines the intelligent interpretation level,the accuracy of image products,and ability of application services.Before processing and application,effective evaluation of remote sensing imagery or payload is required.Spatial resolution is an important index to measure the imaging performance of optical remote sensing payload,which is usually detected by three-bar targets or fan-shaped targets deployed in the remote sensing calibration.The three-bar detection method features simplicity and directness.However,due to discrete deployment,the detection results are easily affected by the phase difference factor,with low accuracy of the method.Fan-shaped targets are usually deployed on a two-dimensional plane.Due to the influence of perspective imaging and geometric distortion,the spatial resolution detection method should be carried out under the condition of vertical photography or geometric correction.Thus,during imaging at a large angle,the method cannot be directly employed for resolution detection due to large geometric deformation.Even under the condition of geometric correction,the resolution detection accuracy will be greatly reduced due to edge reduction or pixel mixing effect.Therefore,according to the isotropic characteristics of the three-dimensional spherical surface,this paper proposes to construct a fan-shaped resolution detection target on a spherical surface,which is called a spherical target.It also hopes that the spherical target can be helpful in solving the problems brought by three-bar or fan-shaped radial targets,and can be directly adopted for spatial resolution detection of high-resolution optical imaging loads under non-vertical imaging and nongeometric correction conditions.Methods Firstly,this paper builds a mathematical model of the spherical target and analyzes the design modes of the spherical target under equal and unequal conditions,with deployment strategies of spherical targets being given.Secondly,based on the traditional two-dimensional target image resolution determination method,an ideal spherical target image resolution determination method is given,and the resolution detecting range is analyzed for different types of spherical targets.Thirdly,the imaging characteristics of spherical targets are analyzed.Additionally,based on the building of the strict imaging model,the central projection deformation rule of the spherical target is studied by parameter simulation.The conclusions are obtained as follows.For multiple concentric circles with the center of the fan-shaped target strip as the origin,imaging ellipses on the two-dimensional images have the same eccentricity.The centers of the imaging ellipses are on the same straight line,while the proportional relationship between the long half axes of the ellipses is approximately equal to that of the image resolution.Finally,an image resolution determination method based on directional moving ellipse fitting is innovatively proposed for the spherical targets,and technical steps of the method are given.Results and Discussions Under the condition of central projection,images are simulated for the concentric circle of the spherical target(Fig.5).Both the imaging characteristics and ellipse fitting characteristics are analyzed through the simulated data.Simulation results show that the distance ratio error of the long half-axis of the spherical target imaging ellipse is low,and the maximum deviation of the simulated data is not more than 1%(Table 4),which is suitable for spatial resolution detection.By employing the actual spherical target fixed in the China(Songshan)satellite remote sensing calibration field,the edge extraction(Fig.6)and ellipse fitting(Fig.7)process on a unmanned aerial vehicle(UAV)sensor target image is conducted,and then the ellipse eccentricity(Table 5)is calculated.Based on multi-scale ellipse construction and sampling point acquisition(Fig.8),the gray curve of the target strip image obtained by the fixed center ellipse fitting method(Fig.9)is intercepted and drawn and is compared with the intercepted and drawn results obtained by the proposed method(Fig.10).Then,the optimal long half axis size of UAV image target ellipse is analyzed(Table 8),and the spatial resolution of UAV imaging load is calculated and evaluated to verify the validity and accuracy of the proposed method.Conclusions This paper aims at improving the imaging performance evaluation effectiveness of aerospace remote sensing payloads.Based on analyzing the advantages and disadvantages of the two-dimensional fan-shaped target in the remote sensing calibration field,it proposed a method for determining the spatial resolution of optical imaging payload directly on the spherical target image.Then,it also conducts experiments on both simulated data and actual UAV spherical target images and verifies the feasibility and correctness of the proposed method for remote sensing payload under tilt imaging conditions.Compared with the traditional two-dimensional fan-shaped target,the proposed method avoids the image correction processing based on ground control points or digital elevation model data and does not need the image geometric positioning model to be involved in the correction solution.In addition,the spherical target image can be directly employed to determine the load imaging performance,with a more accurate spatial'resolution determination effect.The methods and experimental results in this paper can provide technical support for the construction of the spherical target in the remote sensing calibration field and the spatial resolution determination of high-resolution optical remote sensing payload.