Applied Research In Remote Sensing Image Segmentation Technique To Automatically Extract The Feature Information

Sometimes information of land features should be extracted for analysis and decision in next step work during the land resources survey or land-use remote sensing work.The most important issue is how to extract and ensure the quality for the worker.Now the way we use is manual work.But the data is so huge that traditional manual visual interpretation was time and labor consuming.Image segmentation is a key technology that is the basic of digital image processing.The image we need can be extracted.The article focuses on image segmentation applying to remote sensing image in the automatic extraction of land type information.Extracting area feature from remote sensing image belongs to the field of computer vision,image segmentation ie,From the mid-century there are three methods to solve these problems: Image segmentation bases on threshold;Image segmentation bases on edge detection;Image segmentation bases on area.Given some of the characteristics of remote sensing images,such as attributes are similar,pixel is concentrating and area features are facet,combined the character of image segmentation,Otsu ? Canny and Watershed are selected as the methods of this article because Otsu and Watershed are fit for the area and Canny is fit for clear-cut features.The order is as much as possible extracting feature information process from the remote sensing image with small error and gradually analysis on the whole process.What happens is this: Using Visual c + + 2008 development tool as software development platform,debugging success three algorithm program,selecting typical remote sensing image as the experimental images.First the test images were handled by these three ways,the result was analyzed.Second the test images were treated for Grey value,image enhancement and image denoising by the image software and then studied.Third based on the three ways characters,the test images were spitted with the methods of The three algorithms combining or Sequencing,then the test result was analyzed.This paper summarizes through the experiment content and experiment effect.We may safely arrive at the conclusion that Otsu and Watershed solved the features auto extracted in the remote sensing images,and the result is satisfactory.But because of the More complex noise,Canny does not apply in the automatic extraction of remote sensingimage feature information.Before the image segmentation,original image should be preprocessed or the method of combination algorithm may be make a positive impact on automatic extraction of image feature information.Experiments can also be seen by automatic extraction methods in general,better than manual interpretation interpretation extraction,extraction and semi-automatic extraction of human-computer interactive.In this paper,the theoretical analysis and experimental verification support the viewpoint of this article,and achieved the expected effect.Achieve a remote sensing image feature information automatically extracted in land resources survey and land use dynamic monitoring of remote sensing work,can greatly liberated productivity,shorter working hours,lower operating costs,can greatly improve efficiency.