【作者】 林兆华；

【导师】 陈涛；

【作者基本信息】 中国科学院研究生院（长春光学精密机械与物理研究所） ， 机械电子工程， 2012， 博士

【摘要】 光电经纬仪对目标进行跟踪测量过程中，目标与设备的相对距离不断发生变化，会造成目标像面离焦，降低了背景与目标成像的对比度，影响了成像质量。现有的光电经纬仪的调焦是依靠人眼观察手动调焦,或根据距离信息来实现距离调焦，这种传统调焦方式具有精度不高等缺点。同时由于受温度变化使得经纬仪焦面发生变化，为了减小环境温度变化对经纬仪成像系统的影响，目前主要利用温度经验公式补偿温度引起的离焦量，况且每台经纬仪都需要进行外场温度补偿标定，工作量很大。因此自动调焦问题是光电经纬仪亟待解决的问题。随着现代计算技术的发展和数字图像处理理论的日益成熟，越来越多的自动调焦算法运用的是图像处理理论。基于图像处理自动调焦技术在光电经纬仪等大型光测设备中是非常具有应用前景的，在此背景下本文对基于图像处理理论的自动调焦技术进行了相关研究。本文首先根据图像处理自动调焦的原理及其关键技术，给出了基于图像处理的自动调焦技术在经纬仪中的应用的系统组成和工作流程，对目前基于图像处理的清晰度评价函数进行了分类阐述，并且对某些评价函数进行改进，选取几种代表性调焦评价函数进行计算与实验，得到并分析其函数曲线，选出最优的清晰度评价函数。本文将经纬仪自动调焦过程分为粗调和细调两个过程，通过实验得到本文提到的改进的Kirsch清晰度评价函数适合作为粗调清晰度评价函数，基于小波变换的清晰度评价函数适合作为细调清晰度评价函数。针对小波变换清晰度评价函数计算量大，实时性不高的特点，本文采用提升小波变换的清晰度评价函数，采用提升方法构造小波函数，并提出基于提升小波变换的清晰度评价函数。本文提出了一种经纬仪自动调焦窗口的选择方法——基于目标脱靶量计算调焦窗口，该方法能跟踪运动目标，计算量较小，满足调焦准确性的要求，另外还减少了背景对调焦过程的影响。在聚焦搜索过程中，本文提出一种适合经纬仪自动调焦的搜索算法——爬山搜索与曲线拟合相结合的搜索算法，在粗调焦过程中，使用爬山搜索算法，在细调区内，采用爬山算法与曲线拟合相结合的调焦方式。本文对爬山搜索算法存在的问题进行多项改进，实验证明了该方法的有效性。最后对基于图像处理的自动调焦系统在经纬仪上进行了验证，实验表明本文算法能够做到对经纬仪图像的自动调焦，系统的调焦精度为±0.015mm。稳定性较高，该自动调焦系统完全能够满足光电经纬仪实时跟踪以及事后数据处理和分析对图像清晰度的要求。更多还原

【Abstract】 During the tracking measurement process of photoelectric theodolite to target,the relative distance between target and equipment and the ambient temperature arealmost changing, leading to defocus of target image surfaces. It would decrease thecontrast between background and target images, which affects the quality of imaging.The focusing of existing photoelectric theodolite all depends on the observation ofhuman eye and manual focus, or achieving distance focusing according to theinformation of distance. However, this traditional method of focusing takes thedisadvantage of low precision. Meanwhile, the theodolite focal surfaces alwayschanged due to the variation of temperature. In order to reduce the effect of ambienttemperature on the theodolite image system, effort has been done to compensate thedefocusing amount caused by the change of temperature through the empiricaltemperature formula. Every theodolite needs to be compensation demarcated byexternal field temperature. The workload is very huge. Thus, the auto-focusingproblem of photoelectric is urgent to be resolved.With development of the modern calculation technologies and fast maturity ofthe digital image treatment theory, more and more image treatment theory is appliedin auto-focusing algorithm. The application of auto-focusing technique based image treatment in large photic measure equipment, such as theodolite, possesses hugechallenge and good application prospect. In this work, we investigated the applicationof the auto-focusing technique based image treatment in photoelectric theodolite.In this work, we first offer the system composition and process of the applicationof the auto-focusing technique based image treatment in photoelectric theodoliteaccording to the principle and key technology of image treatment auto-focusing. Thenwe stated in classify the current sharpness evaluation function based on the imagetreatment and improved some evaluation function. Moreover, we process someselected representative focus criteria function through calculation and experiment toget and investigate the function curves, consequently, obtained the optimal sharpnessevaluation function. In this paper, the focusing process of theodolite was classifiedinto coarse adjustment and fine adjustment. It is confirmed that the improved Kirschsharpness evaluation function is suitable for using as the coarse adjustment sharpnessevaluation function, and the sharpness evaluation function based on wavelet transformis suitable for using as the fine adjustment sharpness evaluation function.During the calculation on wavelet transform sharpness evaluation function, weinvestigated the selection of the wavelet base through experiment and selected thesuitable wavelet function. Because the characteristics of large calculation and lowreal-time performance of the wavelet transform sharpness evaluation function, weadopted the lifting wavelet transform sharpness evaluation function and used thelifting method to construct the wavelet function. Moreover, we put forward thesharpness evaluation function based on lifting wavelet transform.In this paper, we put forward the selected method of theodolite auto-focusingwindow, which is based on the target miss distance calculation focusing window. Thismethod could track moving target and possess small computational complexity, whichcould satisfy the requirement of focusing accuracy and reduce the effect ofbackground on the focusing process.We also put forward a searching algorithm which is suitable for theodoliteauto-focusing. The searching algorithm is hill climbing search combined with curve fitting. The hill climbing search algorithm is used during the process of coarsefocusing and the focusing mode of hill climbing search combined with curve fitting isused during the process of fine focusing. Because the hill climbing search algorithmexisting a lot of problems, we improved many items in this paper. The effectiveness ofthis algorithm was confirmed by experiment.Finally, we explain the hardware and software design of the image treatmentauto-focusing system clearly. Moreover, the technique was confirmed throughexperiment. The experiment indicates that the theodolite image could auto-focus byusing this algorithm, and the focusing accuracy is±0.015mm. this auto-focusingsystem possesses higher focusing accuracy and could totally satisfy the requirement ofthe photoelectric theodolite real-time tracking, data treatment and analysis afterwardsfor image definition.更多还原