Remote sensing image cloud detection method based on pseudo color and support vector machine

Abstract

The invention discloses a remote sensing image cloud detection method based on pseudo color and a support vector machine, and is mainly to solve the problems of low precision and large computation amount in the prior art. The method has the following realization steps: 1) selecting three wavebands within intermediate infrared and thermal infrared band range from the multiple wavebands of remote sensing data, and with the image grey values of the three wavebands being as RGB components, synthesizing pseudo-color images; 2) with RGB vectors of training images and cloud and non-cloud priori categories being as input, calculating the optimal classification hyperplane of the characteristic space by utilizing a support vector machine method with a kernel function to construct a classification decision function; and 3) for an image to be detected, with the image grey values of the three wavebands being as the RGB vectors, carrying out cloud detection through calculating the value of the decision function. The method provides the decision function for the remote sensing image cloud detection through training, thereby not only improving precision of the remote sensing image cloud detection, but also reducing the computation amount; and the method can be used for intermediate infrared and thermal infrared band range remote sensing image detection.

Description

technical field [0001] The invention belongs to the field of aerospace remote sensing and relates to a cloud detection method for remote sensing images, which is suitable for detecting remote sensing images including mid-infrared and thermal infrared bands. Background technique [0002] The ground measurement and control stations of my country's remote sensing satellites are basically located within the limited territory of our country, and the satellites can only transmit image data to the ground within a limited range of steps passing through the measurement and control stations. In the process of remote sensing data transmission, there is a contradiction between large amount of data and short transmission time. At present, the main method to solve this problem is to use image compression technology to reduce the amount of data, but the amount of data that image compression technology can reduce is also limited. Excessive compression will cause the loss of image detail inf...

AI Technical Summary

Problems solved by technology

This type of method can only be performed on block images, not suitable for single pixels, requires a large amount of memory, and the algorithm is complex, requiring a faster processor

It is mainly used for cloud detection by ground remote sensing staff, which is difficult to achieve in the limited hardware environment on the star

[0007] The above-mentioned first type of method needs to extract multiple threshold parameters, which is completed by manual experience and lacks a general system method. It is difficult to extract parameters in the case of a large number of spectral bands, and the reliability of the detection result is not high; the above-mentioned second type of method needs to consider the ever-changing The shape, texture and other characteristics of the cloud, the amount of calculation is large

Images