降低无人机影像数据冗余度方法的研究

发布时间：2019-01-03 07:07

【摘要】：无人机影像重叠度较高,幅数过多,影像拼接需要耗费大量时间,在不影响数据质量的情况下,如何降低航片的冗余是本论文研究的主要问题。论文以甘肃某县土地确权项目为依托,以实测无人机影像为基础数据,对减小无人机影像重叠度,减少数据冗余进行了分析研究,具体研究内容及结果如下:1.首先对研究区域无人机数据做了影像重叠度统计,结合无人机最佳重叠度和规范要求,发现无人机影像存在大量冗余。针对无人机影像高重叠度的特点,采用不同间隔的抽稀方式对影像稀释。实验结果表明影像抽稀间隔越大,拼接影像所用时间越少,平面位置中误差越大。间隔4张影像抽稀达到影像重叠度规范临界值,拼接的时间比未抽稀影像拼接所用时间减少了一半多。2.通过构建一个包含138幅航空遥感影像的样本库,利用知识库的方法,在10分制的基础上,提出了利用加权平均标准差、平均梯度和信息熵三种评价指标得到影像质量的综合评价值,建立一种基于综合评价值的抽稀方式。其影像拼接比全部影像拼接短了近一半时间,融合影像平面位置中误差得到的良好控制,曝光过度影像引起的“白斑”现象明显变少,影像辨识度明显提高。3.影像裁剪也降低影像重叠度,加快无人机影像影像匹配速度。同时,无人机影像影像边缘畸变较大,裁剪后的影像边缘基本被剪掉,可以大大提高影像的配准精度。针对影像裁剪的这一特点,提出影像裁剪和抽稀相结合的方式对影像进行处理,进一步缩短影像拼接用时,改善了影像边缘畸变对影像拼接成图的影响。
[Abstract]:Unmanned aerial vehicle (UAV) images have a high degree of overlap, too many amplitudes, and it takes a lot of time to concatenate images. Without affecting the quality of data, how to reduce the redundancy of aerial photographs is the main problem in this paper. Based on the land right confirmation project of a county in Gansu province and the measured UAV image as the basic data, this paper analyzes and researches on reducing the overlap degree and data redundancy of UAV image. The specific research contents and results are as follows: 1. Firstly, the image overlap degree of UAV data in the study area is calculated. Combined with the optimal overlap degree and specification requirements of UAV, it is found that there is a lot of redundancy in UAV images. Aiming at the characteristics of high overlap degree of UAV images, thinning methods with different intervals are used to dilute the images. The experimental results show that the larger the thinning interval is, the less time is used for image stitching and the greater the error in plane position is. At intervals of 4 images, the critical value of image overlap degree is reached, and the time of stitching is more than half of that of unextracted images. By constructing a sample database containing 138 aerial remote sensing images and using the method of knowledge base, on the basis of 10 points system, the weighted mean standard deviation (WMSD) is put forward. The average gradient and information entropy are used to evaluate the image quality, and a thinning method based on the comprehensive evaluation value is established. Its image stitching time is nearly half shorter than that of all image stitching, and the "white spot" phenomenon caused by overexposure image is obviously reduced, and the recognition degree of image is improved obviously, and the good control of the error in the plane position of the image is obtained. Image clipping also reduces image overlap and speeds up UAV image matching. At the same time, the edge distortion of UAV image is large, and the edge of the cut image is basically cut off, which can greatly improve the registration accuracy of the image. In view of this feature of image clipping, the combination of image clipping and rarefaction is proposed to further shorten the image stitching time and improve the influence of image edge distortion on image mosaic.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P237

【参考文献】