摘要
与条形码、标准答题卡相比较,惯性导航信息标记点的图像有许多不可预测的污染,比如曝光质量差、灰尘、指纹以及其他形式的污迹;另外,扫描分辨率不同、参考标记点的起始位置不确定也会影响标记点的正确识别与定位。提出了一种基于聚类的一维投影波形分析算法,并成功应用于航拍胶片数字图像上的惯性导航信息标记点的自动识别技术中。在一定程度上解决了在噪声不可控的情况下进行标记点识别定位问题。通过统计方法,降低了噪声对结果的影响;通过把投影转换成一维信号的概念,提供了一致的分析方法,提高了处理效率;通过引入聚类思想计算标记点高度、宽度以及标记点的间距,大大提高了算法的稳定性。
There is much unpredictable pollution on the inertial navigation information marker point＇s images comparing with the bar code, the standard answer cards, for example, the exposure of poor quality, dust, fingerprints and other forms of smudged. In addition, the scanning resolution, the uncertainty of the beginning position of the standard reference marker points will also affect identifica- tion and location correctly. This paper presents a cluster based on the one-dimensional projection waveform analysis algorithm, and successfully applied to the film Aerial digital images on the inertial navigation information markers of automatic identification technology. To some extent, to solve the noise is not controllable manner markers to identify the issue of position. By statistical methods, we project the binary image of the inertial navigation information marker point＇s images in the X-axis and Y-axis. Based on the projection result and the statistic data, we do some analyse and processing. It is very useful to reduce the noise impact on the results. Through converting the projection data into the concept of one-dimensional signals, providing a consistent analytical method, the impact of the noisy points can be detected and removed. At the same time the omitted marker points can be found and added to the correct result. It is very clear that this method will enhance the efficiency of the processing and increase the rate of marker point identification correctly. During the analyse processing, we introduce the ideal of clustering to calculate the points＂ height, points＂ width and the spacing width between points＇, which has greatly enhanced the stability of the algorithm. In addition, the paper also takes an image as an example to illustrate how the method presented in this paper is used to analyse the experimental data. And present a detailed step to calculate the wanted result. Finally, the performance of the analysis method is discussed. The performance tests experiments show that the method in practical applications achieved good results.
出处
《重庆师范大学学报（自然科学版）》
CAS
2007年第4期58-61,共4页
Journal of Chongqing Normal University:Natural Science
关键词
航拍图像
惯性导航信息
自动识别
聚类
aviation image
inertia navigation information
automatic recognition
clustering
