论文部分内容阅读
为实现全球环境实时、动态监测的需要，选用Ｆｏｒｓｔｎｅｒ特征匹配理论对ＮＯＡＡ／ＡＶＨＲＲ影像进行匹配试验。Ｆｏｒｓｔｎｅｒ特征提取之前，先根据熵的理论对影像进行预处理．特征提取工作量可减少到１０％～２０％。特征提取后的共轭影像，根据相似性尺度可建立一个可能的同名点匹配点表，再根据一致性尺度，采用Ｒｏｂｕｓｔ稳健估计剔除错匹配的点对，即可建立两幅影像的最终匹配点对。匹配结果表明，通过阈值的设置，可以在ＡＶＨＲＲ多时相影像上提取足够数量的匹配点对，在行、列方向上平均误差约为０．６像素和０．３像素。
To meet the need of real-time and dynamic monitoring of the global environment, Forstner’s feature matching theory was used to match NOAA / AVHRR images. Before the Forstner feature extraction, the image is pre-processed according to the entropy theory. Feature extraction workload can be reduced to 10% to 20%. According to the similarity scale, a possible matching point list with the same name may be established. Based on the consistency measure, Robust robust estimation is used to reject the wrong matching point pairs to establish the final matching point of the two images Correct. Matching results show that by setting the threshold, a sufficient number of matching point pairs can be extracted from the multi-temporal AVHRR images and the average error in the row and column directions is about 0.6 pixel and 0.3 pixel.