Thick Cloud Detection In Ocean Remote Images And Study Of Fusion In The FFT Domain

Ocean remote sensing technology is one of the important research contents of global environmental monitoring. Environmental monitoring is not only the monitoring of the ecological environment, but also related to the security environment. The work of marine target detection is of great significance. How to detect the marine target effectively from the continuous development and progress of the satellite technology,will be one of the key technologies of the research. In order to achieve this goal, a large number of works and studies were carried out in our research group. At present, the major works has completed are “Accurate registration of remote sensing images” and “Sea-land segmentation of the remote sensing image”. The works of this paper is first detect the thick cloud from the remains ocean region after land-sea segmentation, then Study on image fusion and dimension reduction techniques for remote sensing images. The work of thick cloud detection and sea-land segmentation have the same purpose, namely exclude interference regions. And study on dimension reduction is in order to reduce data and use the data more effectively. Finally, the target will be detected on the ocean area of remote sensing image after dimensionality reduction.The main work and research contents of this paper are as follows:1) By analyzing the gray characteristics of cloud and sea images at different tower levels,we find that ocean in different regions have different distribution characteristics, and ocean in the same region always have the same distribution characteristics. But thick cloud always has different distribution even they are in the same area. Based on those conclusions, we construct the tower structure of sea area, and then establish gauss background modeling at the top of the tower. Using the established model to detect the thick cloud of the ocean sub block can reduce the interference of sea clutter and improve the efficiency of detection.2) Remote sensing image fusion in Fourier domain is analyzed and studied, the Fourier domain fusion selected suitable amplitude spectrum and phase spectrum image from the transform component after Fourier transform in the spectral dimension of remote sensing image, and then compose the color image in HSV color space, then transformation HSV space mapping to RGB color space, the final color fusion image is obtained. This paper presents the methods of Multi-spectral Remote Sensing Image fusion in Fourier domain and Multi-spectral and high resolution Remote Sensing Image fusion in Fourier domain, and introduces the principle and process of the algorithm. Finally, by contrasting the Fourier domain fusion algorithm to other fusion algorithms, the effectiveness of the fusion algorithm is verified.3) By analyzing the characteristics of different components of the spectral dimension Fourier transform, it is found that the second harmonics of the spectral dimension Fourier transform, has the similar physical expression and physical meaning of a color point in HSV color space, DC component and the luminance component have similar physical meaning. These results indicate that the fusion of Fourier domain is consistent with the visual characteristics of the human eye, and it also constitutes the physical basis for the dimensionality reduction technique of this paper. Dimensionality reduction method based on Fourier domain fusion, can make full use of information of every spectral and high resolution Remote images, and will not cause data leakage, and in a certain extent can suppress the sea clutter.The work of this paper is of great value for the detection of ocean targets in remote sensing images. Thick cloud detection can reduce the interference caused by other things when the sea target detection work is under. Remote sensing image fusion and dimension reduction can extract information from the large amount of remote sensing data.