环境卫星多光谱相机CCD波段间配准精度评价研究

发布时间：2018-05-21 05:12

本文选题：多光谱相机 + 拟合算法　；参考：《辽宁工程技术大学》2013年硕士论文

【摘要】：多光谱相机可获取多个波段的影像,通过后续的配准、融合等处理能够生成多种专题影像产品,极大地丰富了影像信息同时也提高了影像数据的应用潜力。多光谱相机已成为当前遥感卫星搭载的重要载荷之一。影响CCD相机波段配准的因素有很多,比如温度变化和结构稳定性等。波段间配准是多光谱影像处理的重要环节,配准精度直接影响影像数据后续处理和应用价值。由于高程起伏、传感器内部畸变等因素的影响,卫星多光谱相机获取的影像数据波段间通常可能存在较大的非线性几何变形。因此通过简单的面平移和旋转,去实现多光谱影像的波段间配准是无法满足产品质量的精度要求。对于高空间分辨率卫星来说,波段间配准的精度对定位精度、地物分类准确性与可应用性的影响尤为显著。本文在结合各传统的遥感图像配准的基础上,针对我国环境卫星的特点,对环境卫星多光谱相机四个CCD波段的装配、运输及发射过程中产生的误差进行评价,最终达到各波段间配准误差的准确表达的目的。本文在配准算法和亚像元定位技术的基础上,提出了几种专门针对环境卫星特性的波段配准精度评价的算法。其中包括灰度质心算法、高斯曲线拟合算法、霍夫变换交点算法、高斯曲面拟合算法以及基于特征地物评价算法。这些算法分别从影像的灰度结构、边缘曲线的性质、局部曲面的性质以及影像特征区域灰度的分布特性等方面对影像的配准误差进行了详细的评述。基于这些算法,本文计算了环境卫星多光谱相机CCD波段间配准误差分别在沿轨和跨轨方向上的数值大小。计算结果显示环境CCD二级产品各波段之间的配准误差基本小于0.5个像素,部分数据可以达到0.1个像素之内。几种算法都可以获得比较稳定的评价结果,从而达到了从多方面对波段间配准误差的评价,相互验证了计算结果的证确性。
[Abstract]:Multi-spectral camera can obtain multiple bands of images. Through subsequent registration, fusion and other processing can generate a variety of thematic image products, greatly enriched the image information, but also improved the application potential of image data. Multispectral camera has become one of the important payloads of remote sensing satellite. There are many factors affecting CCD camera band registration, such as temperature variation and structural stability. Inter-band registration is an important part of multi-spectral image processing. Registration accuracy directly affects the image data processing and application value. Due to the influence of height fluctuation and internal distortion of sensor, there may be large nonlinear geometric deformation between the bands of image data obtained by satellite multispectral camera. Therefore, it can not meet the precision requirement of product quality by simple surface translation and rotation to achieve the registration of multi-spectral images between bands. For high spatial resolution satellites, the accuracy of registration between bands has a significant effect on positioning accuracy, classification accuracy and applicability of ground objects. Based on the traditional remote sensing image registration and the characteristics of Chinese environmental satellites, this paper evaluates the errors in the process of assembling, transporting and launching four CCD bands of environmental satellite multispectral cameras. Finally, the accurate expression of registration error between different bands is achieved. On the basis of registration algorithm and sub-pixel location technology, this paper presents several algorithms for evaluating the band registration accuracy of environmental satellites. It includes gray centroid algorithm, Gao Si curve fitting algorithm, Hough transform intersection algorithm, Gao Si surface fitting algorithm and feature based feature evaluation algorithm. In this paper, the image registration errors are reviewed in detail from the aspects of the gray structure of the image, the properties of the edge curve, the properties of the local surface and the distribution of the gray level in the feature region of the image. Based on these algorithms, the magnitude of the registration error between CCD bands of environmental satellite multispectral camera is calculated in the direction of orbit and trans-orbit, respectively. The results show that the registration error is less than 0.5 pixels and part of the data can be within 0.1 pixels. Several algorithms can obtain relatively stable evaluation results, which can evaluate the registration error between bands from many aspects, and verify the validity of the calculation results.
【学位授予单位】：辽宁工程技术大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P237

【参考文献】