The Algorithm For Identification Of Island Based On Optical Remote Sensing

The sea islands link national maritime security and maritime rights and interests. The survey for sea islands is essential for national maritime security and the management of maritime space. Traditionally, survey for sea islands was quite a laborious and time-consuming process. Remote Sensing (Rs) is an efficient way to survey for the locations, shapes, areas, and coastline lengths, and the exploitation and utilization of sea islands. Aimed at Automatic Island Identification, this paper presents an improving algorithm for island accurate geometric correction, pixel gray-scale re-sampling, island polygons extraction by image binarization and integration of Rs, Gis and DEM data, in view of the problems of low precision of geometric correction, slow interpretation speed, heavy workload and low automatic level. After analyzing the cause of inaccurate statistics of the number of islands in Zhajiang Province, statistics rules have been established and applied to get the number, areas and deep-water shorelines of the islands in Zhajiang Province accurately. The main contributions of this paper are as follows:1. An improving algorithm named pixel correlation method for accurate geometric correction has been first proposed. With consideration of the fact that remote sensing data could only provides a few ground control points with low accuracy level of islands. Geometric correction includes the following steps: Firstly, the model between multi-spectral images and projection on the ground coordinate system should be established, as well as the correlation model between multi-spectral image pixel coordinates and pan image pixel coordinates. Secondly, the accurate geometric correction of multi-spectral images be processed. Then, based on the correlation model, the accurate geometric correction of pan images could be done. The improving algorithm solves the double-fold phenomenon effectively caused by insufficient amount of ground control points under the circumstances of oceanic environment. It is also suitable for land.2. A Lagrange isoperimetric element interpolation algorithm and detailed the formula process are proposed. The result proved that 4-node 1-order Lagrange isoperimetric element interpolation is equivalent to bilinear interpolation, while 16-node 3-order to cubic convolution. 12 pictures have been selected to evaluate the accuracy of the algorithm, including PSNR, bias index, correlation coefficient, warping degree and MSE. The result of all the image quality assessment index proved that Lagrange isoperimetric element interpolation algorithm is neither inferior to bilinear interpolation, nor to cubic convolution.Constraint condition of interpolation was led in to avoid serious deviation between the calculation value and the true value, which has been proved that the image quality can be improved significantly.The cause of 9-node 2-order interpolation error has been analyzed. The study proposed a mixed interpolation, which means that different mixture interpolation algorithm should be adapted according to the sampling position.3. R-Factor analysis has been applied to separate the superposition information in ETM+ image. By rotating the main factor, the main index is extracted which indicate the features of the islands. Processing the main index by bi-value with dynamic threshold, bi-value image could be formed. Calculating the bi-value images of different ETM+, the moving target could be cleared and the polygons of island would show up. It is the theoretical basis of the island identification.4. The paper proposed a algorithm based on Rs, Gis and DEM data, which fills up the blank of mapping techniques applying the integration of Rs, Gis and DEM data.5. The paper established statistics rules of the number and area of islands in Zhejiang province. Utilizing multi-time TM and ETM+, as well as SPOT5 and oceanic maps, the islands could be identified. To obtain the information of the number and area of islands has important significance to protect the ecological system and rationally exploit and utilize the resources of islands.6. An algorithm was proposed to identify the deep-water Shoreline based on the Rs, Gis , DEM data and the shoreline. It has great meaning for the port planning and construction in Zhejiang province.