Research On Remote Sensing Image Segmentation And Monitoring Method Of Surface Water Based On Morphology

In the process of water resources protection,statisticing the area of water bodies is one of the main steps to understand the basic conditions of the water resources environment,and also provides data support for the protection of water bodies.The traditional water area survey is based on artificial measurement to the field.But manual survey requires a lot of manpower and material resources,and the rate and cycle of measurement is low and long.With the high-resolution satellites No.1 and No.2 launching successively in China and the matyrity of high-resolution related technoligies,remote sensing technology has evolved from the qualitative analysis of data to the the quantitative analysis,so that digital image methods can be widely used in processing remote sensing images.And then this method replaces the manual method,has solved the problems of long measurement period and low efficiency.Therefore,this paper proposed a method for measuring water area using remote sensing images of surface water bodies.The main work of this papper is(1)Firstly,this papper analyzes the domestic and foreign research status of segmentating the water area in remote sensing image and the basic principles and introduces the image processing methods which using morphological.Three kinds of methods segmentating water area are introduced in detail: threshold-based image segmentation,edge detection-based image segmentation and region-based segmentation methods.These methods are simulated to analyze their advantages and disadvantages.(2)Image has the weak edge strength and unclear contour,these features lead to segmenting image difficultly.Aiming at the above questions,an edge detection method based on morphology multi-scale adaptive equalization is proposed.Firstly,the morphological reconstruction and gradient calculation of grayscale remote sensing images are performed by using four scales of circular structural elements and linear structural elements in four directions respectively,and multiple images with different scale and direction edge features are obtained.Secondly we compute the weight ratio of grey value variance of each image,Self-adaptive summation of multiple images using weight ratio to obtain edge contours.The experimental results show that,Compared with classical edge detection operators,the edge sharpness(average gradient)and the amount of edge information contained in the image(entropy of information)are used as evaluation indicators to show that the edge detection method of this paper is superior to traditional operators and has lower computation time.It is easy to introduce noise in the collection,transmission and decoding of remote sensing images.The signal-to-noise ratio of the method is analyzed.The comparison shows that the edge detection method has strong anti-noise performance.(3)Aiming at the problem of low segmentation accuracy and complex method flow in traditional single segmentation methods,a segmentation method of surface water body segmentation based on edge extraction of mathematical morphology and region growth was proposed.In the Matlab7.0 environment,the pixel area of the segmented water was counted and compared with the area of the hand-painted water body.Simulation experiments show that the proposed method has a segmentation error of approximately 3.5%,which is superior to the marker-controlled watershed algorithm and the K-means-based segmentation algorithm,and the computational speed is relatively fast.