高速相机自动快速调光调焦方法研究

发布时间：2018-06-10 08:31

本文选题：高速相机 + 自动调光　；参考：《中国科学院长春光学精密机械与物理研究所》2017年博士论文

【摘要】：随着靶场需求的不断提高,光电经纬仪所捕获跟踪目标的速度越来越快,目标相对于经纬仪的位置以及背景光照变化也越来越剧烈,对光电经纬仪的调光调焦精度和时间要求也越来越高。为了捕获与跟踪高速运动目标,获取瞬间变化现象的细节,现代光电经纬仪多采用高速相机作为图像采集设备。为了获得高清晰度图像,必须解决高速相机自动快速调光调焦问题,以适应高速目标捕获跟踪的需求。因此,快速自动的调光调焦方法研究成为当前研究的焦点和热点。目前市面提供的高速相机鲜有具备自动调光调焦功能,而传统光电经纬仪的自动调光调焦方法精度较低,速度慢,不能适应对高速运动目标捕获跟踪需求。本文针对这一问题,面对高速相机的特点,提出了基于快速图像处理的、适用于高速相机的自动快速调光调焦方法,对调光调焦原理进行了深入研究,研究内容及创新性工作主要如下:第一,本文建立了相机与光照强度的数学模型,通过分析相机的过曝光及欠曝光产生的原理及光电经纬仪用高速相机在工作时需要应对的光照环境,提出了一种基于图像直方图特征函数(Histogram Feature,HF)的图像亮度评价方法。与传统调光评价方法相比,该方法在含有光强大幅度变化的顺光、逆光两种光照条件下对图像评判均有较好的效果,且通过提取兴趣区域信息降低计算量后,该函数克服传统图像质量评价函数计算时间普遍较长的问题,计算时间小于0.2ms,为了进一步提高函数计算的实时性,对该函数的硬件加速性能进行了研究,实验表明基于GPU的硬件加速方法在图像分辨率较高时加速效果明显;第二,针对传统搜索方法进行了分析对比,并通过运用图像直方图特征函数,提出了一种曝光粗调与精调相结合的自动调光方法,并通过静态实验验证了方法相较于传统方法,在应对110ms时间内光照变化范围约为30倍时的场景时,其捕获图像序列的信息熵均值可以提高约2.648,方差比传统方法下降1.125。第三,通过运用光学系统的成像数学模型,对离焦深度法中点扩散函数及其调焦方法与对焦深度法中常用的图像质量评价函数进行了分析。提出了一种用于自动调焦的基于图像直方图特征函数的图像评价函数。实验证明,该函数可以克服传统自动调光调焦评价函数需要分开计算重复占用系统资源,以及计算速度较慢的缺点;并且该函数的精度以及调焦搜索范围在各类函数中均处于适中的水平,可以满足高速相机自动调焦的需求。第四,通过图像直方图特征函数进行分析,提出了一种自动调焦方法,实验证明,与传统搜索方法相比,快速方法在保证调焦精度的同时,可以有效减少电机的往返运动次数,对于距离准焦位置70步外的目标,调焦电机可以在9次移动内完成调焦并且调焦电机仅需改变方向一次;第五,为了验证光电经纬仪在实际工作中需要应对的情况,针对本文提出的自动调光及自动调焦方法,为了模拟光电经纬仪在实际工作中需要应对的情况,给出了将调光、调焦联合在一起的方法流程,并设计了验证实验,实验证明在调光调焦联合实验中与应用传统搜索自动调焦方法的实验相比,应用快速自动调焦方法的调光调焦联合实验也能找到正确的准焦位置,并且仅耗时不到0.9s,为实际工程应用提供借鉴方案。最后,对全文所做的工作进行了总结,提出了今后需要继续研究的问题和研究方向。
[Abstract]:With the increasing demand of the range, the speed of the tracking target captured by the photoelectric theodolite is getting faster and faster. The target relative to the position of the theodolite and the background light change are becoming more and more intense. The precision and time requirement of the photoelectric theodolite are also getting higher and higher. In order to capture and track the high-speed moving target, the instant change is obtained. In detail, modern photoelectric theodolite mostly uses high-speed camera as image acquisition equipment. In order to obtain high definition image, it is necessary to solve the problem of automatic and fast dimming and focusing of high-speed camera, so as to meet the demand of high speed target capture and tracking. The high speed camera provided in the market has the function of automatic dimming and focusing, and the traditional photoelectric theodolite's automatic dimming and focusing method has low precision and slow speed. It can't adapt to the requirement of tracking the high speed moving target. In this paper, a fast image processing based on the characteristics of high-speed camera is proposed, which is suitable for high speed phase. The automatic and fast dimming and focusing method of the machine is studied in detail. The main research contents and innovative work are as follows: first, the mathematical model of the camera and light intensity is set up in this paper. The principle of overexposure and under exposure of the camera and the need to respond to the high speed camera of the photoelectric theodolite in the work are analyzed. In light environment, an image luminance evaluation method based on Histogram Feature (HF) is proposed. Compared with the traditional light modulation evaluation method, the method has good effect on the image evaluation under two illumination conditions, which contain a large amplitude of light intensity and a backlight, and the information is reduced by extracting interest area. After the calculation, the function overcomes the problem that the traditional image quality evaluation function has a long time to calculate, and the calculation time is less than 0.2ms. In order to further improve the real-time performance of the function calculation, the hardware acceleration performance of the function is studied. The experiment shows that the speed method based on the hardware of the GPU is more effective when the image resolution is high. Second, according to the analysis and comparison of the traditional search method, and by using the image histogram feature function, an automatic dimming method combining the exposure rough and the fine phase is proposed. And the static experiment proves that the method is captured when the light variation range is about 30 times when the 110ms time is about 30 times compared to the traditional method. The mean entropy of the image sequence can be improved by about 2.648, and the variance is reduced by 1.125. third than the traditional method. By using the imaging mathematical model of the optical system, the point diffusion function and its focus method in the defocus depth method and the image quality evaluation function commonly used in the focus depth method are analyzed. A kind of auto focusing is proposed. An image evaluation function based on the feature function of the image histogram shows that the function can overcome the shortcomings of the traditional automatic dimming and focusing evaluation function, which need to calculate the resource of the system, and the slow speed of calculation, and the accuracy of the function and the focus of the focus search are at the moderate level in all kinds of functions. In order to meet the needs of auto focusing of high speed cameras. Fourth, an automatic focusing method is proposed through the analysis of the image histogram feature function. The experiment shows that, compared with the traditional search method, the fast method can effectively reduce the number of back motion of the motor while guaranteeing the focus accuracy, and for the target with the distance of the para focus 70 steps, Focusing motor can be focused in 9 movement and focusing motor only need to change direction one time; fifth, in order to verify the situation that the photoelectric theodolite needs to deal with in actual work, aiming at the automatic dimming and auto focusing method proposed in this paper, in order to simulate the situation that the photoelectric theodolite needs to deal with in the actual work, the paper gives the method. The process of adjusting the light and focusing together, and designing the verification experiment. The experiment proves that in the light focusing and focusing joint experiment, compared with the traditional search automatic focusing method, the application of the fast auto focusing method can also find the correct para focus position, and it takes only less than 0.9s, which is a practical project. Finally, the paper summarizes the work done in this paper, and points out the problems and research directions that need to be further studied in the future.
【学位授予单位】：中国科学院长春光学精密机械与物理研究所
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP391.41

【相似文献】