推扫式遥感相机对焦评价方法研究

发布时间：2019-01-27 10:52

【摘要】：遥感成像技术在国民经济与国家安全领域起着极为重要的作用,而遥感器作为遥感技术中的核心一直是重点研究的对象。当前,基于TDICCD的推扫式遥感相机作为遥感器的一种,在航天遥感技术中越发显示出其重要性。由于空间环境的影响,遥感相机往往会发生离焦,因而为了获取高质量图像,需要进行有效的自动对焦技术。而推扫式遥感相机由于其成像的特殊性,使得其在任意时刻所拍摄到的景物都不同,从而需要根据其特点提出一种与摄影目标景物无关的对焦评价方法,继而用于推扫式遥感相机自动对焦当中。针对上述背景,本文介绍了TDICCD推扫式遥感相机成像机理,并对目前可用于推扫式遥感相机的对焦评价方法进行了分析,从而在此基础上分别从小波包域与空间域提出了两种与摄影目标景物无关的对焦评价方法——基于小波包能量谱的对焦评价方法与基于强边缘宽度直方图的对焦评价方法。在基于小波包能量谱的方法中,本文先是介绍了小波包技术具有频率分解精细、时频特性良好的技术特点,并借鉴功率谱一致性的理论,发现不同内容图像间在小波包域具有几乎一致的能量谱曲线,从而建立了基于小波包能量谱不变性的观点。算法先是对图像进行小波包分解,接着选取与相机运动方向相垂直的不同频率下的小波包系数,继而对这些小波包系数进行处理得到相应频率下的能量表征,最终对各频率下的能量进行加权求和得到清晰度评价值,从而指导相机调焦。通过仿真实验与实拍实验发现：该对焦评价方法在无像移失配或在有像移失配但离焦模糊不是过大的情况下可做到与摄影目标景物无关的绝对式检焦。在基于强边缘宽度直方图的方法中,本文以遥感图像大多具有直线结构为依据,根据模糊会引起边缘扩散为指导,从而以边缘的扩散情况来表征图像清晰度。算法先是通过边缘检测算子选取与相机运动方向平行的强边缘,由强边缘垂直方向的灰度变化计算其边缘宽度,之后对所有的强边缘宽度建立直方图,继而引入基于直方图信息的距离因子,针对不同宽度的边缘加入不同的权重,最终求和得到清晰度评价值,从而指导相机调焦。通过实拍实验与模糊图像库的测试发现：该对焦评价方法具有与摄影目标景物无关的绝对式检焦特点,对焦曲线拥有良好的单调性,并不受像移失配的干扰,还具有与主观评价一致性的特点,是一种较为有效的推扫式遥感相机对焦评价方法。
[Abstract]:Remote sensing imaging technology plays an important role in the field of national economy and national security, and remote sensor as the core of remote sensing technology has been the focus of research. At present, push-sweep remote sensing camera based on TDICCD, as a kind of remote sensor, has become more and more important in space remote sensing technology. Because of the influence of space environment, remote sensing camera will often defocus, so in order to obtain high quality image, it is necessary to carry out effective automatic focusing technology. Because of the particularity of its imaging, the push-scan remote sensing camera has different scene at any time, so it is necessary to put forward a focusing evaluation method which is independent of the target scene according to its characteristics. It is then used in automatic focusing of push-sweep remote sensing camera. In view of the above background, this paper introduces the imaging mechanism of TDICCD push-scan remote sensing camera, and analyzes the focusing evaluation method which can be used for push-scan remote sensing camera. On the basis of this, two methods of focusing evaluation independent of the object of photography are proposed from the wavelet packet domain and the spatial domain, namely, the focus evaluation method based on wavelet packet energy spectrum and the focus evaluation method based on the strong edge width histogram. In the method based on wavelet packet energy spectrum, this paper first introduces that wavelet packet technology has the characteristics of fine frequency decomposition and good time-frequency characteristic, and draws lessons from the theory of power spectrum consistency. It is found that there are almost consistent energy spectrum curves between different content images in wavelet packet domain, thus the viewpoint of energy spectrum invariance based on wavelet packet is established. The algorithm firstly decomposes the image with wavelet packet, then selects the wavelet packet coefficients at different frequencies perpendicular to the camera motion direction, and then processes these wavelet packet coefficients to obtain the energy representation at the corresponding frequency. Finally, the weighted summation of the energy at each frequency is used to obtain the definition evaluation value, which can guide the camera to focus. Through the simulation experiment and the real film experiment, it is found that the method can achieve absolute focus detection without image shift mismatch or without image shift mismatch, but the defocus blur is not too large, which has nothing to do with the object of photography. In the method based on strong edge width histogram, this paper is based on the fact that most remote sensing images have straight line structure, which is guided by the edge diffusion caused by blur, so that the image clarity can be represented by the edge diffusion. The algorithm firstly selects the strong edge parallel to the camera's moving direction through edge detection operator, calculates the edge width from the gray level change in the vertical direction of the strong edge, and then establishes the histogram for all the strong edge widths. Then the distance factor based on histogram information is introduced, and different weights are added to the edges of different widths, and the definition evaluation value is obtained by summing up, thus guiding the camera focusing. It is found that the method has the characteristics of absolute focus detection independent of the object of photography, the focus curve has a good monotonicity and is not disturbed by the image mismatch, through the experiments of real time shooting and the test of fuzzy image database, the results show that the method of focusing evaluation has the characteristics of absolute focus detection independent of the object of photography, and the focus curve has good monotonicity. It is also consistent with subjective evaluation and is a more effective method for focusing evaluation of push-sweep remote sensing camera.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP73

【参考文献】