The Detection And Display Of The Lake's Historical Change Based On Remote Sensing Technology

In this paper, the image data of Shahu Lake were extracted by different periods of remote sensing image data of Wuhan city, and the image data were classified by multi-scale segmentation and decision tree classification. The mobile terminal was used to display the historical changes of Shahu Lake,It provides a fast, accurate and convenient way for the law enforcement officers of the lake to detect the phenomenon of filling the lake. And improve the efficiency of law enforcement officers, at the same time it also provides evidence for reference.Change detection method used in this paper is image classification comparison, the research is mainly divided into two aspects: remote sensing image preprocessing and object-oriented image classification; Detecting the change area in mobile terminal based on post classification image. The main research contents and innovations include the following:1. the research of remote sensing image preprocessing technology. Remote sensing image preprocessing includes three processes: geometric correction, image mosaic and image cropping. The aerial image data obtained from Geographic Information Center of Hubei Province were processed to obtain the image data of Shahu Lake, and through resampling the image data over the years with the same resolution (1M).2. Optimal segmentation scale in multi-scale segmentation. The multi-scale segmentation method in the object oriented remote sensing image classification, the selection of the scale parameters has great influence on the results. The objects with different sizes and numbers are obtained when the image is segmented with different segmentation scales. In general, the smaller the scale, the more objects will be generated,resulting the over segmentation phenomenon, on the contrary, the larger the scale, the less the object, will lead to Insufficient segmentation.Because of the difference in the spectrum,texture and structure of different object,the different objects have different optimal segmentation scale. The segmentation quality function based on the principle of high homogeneity and high heterogeneity. The optimal segmentation scale is judged by comparing the value of segmentation quality function under different scale parameters.3. Research and improvement based on multi-scale segmentation and decision tree classification. In this paper, the multi-scale segmentation and CART decision tree classification are studied and analyzed. The concept of image block is introduced in the traditional multi-scale segmentation method, and two kinds of block methods are proposed: based on the image "splicing line" and based on large scale multi-scale segmentation. Multi-scale segmentation and decision tree classification for sub block.Taking the image data of Shahu Lake in 1981 as the experimental object, The accuracy of the two improved classification methods is compared with the traditional multi-scale segmentation and decision tree classification. The results show that the accuracy of block-based multi-scale segmentation and decision tree classification is obviously higher than the multi-scale segmentation and decision tree classification for the whole image.Among them, the multi-scale method based on large scale has the highest accuracy.4. Based on the intelligent mobile terminal lake change region detection. This paper briefly introduces the App system of water culture in Wuhan, and the detection of the change area based on classified image is realized in the system change function module.