SAR image classification method based on sparse representation and Gaussian distribution

Abstract

The invention discloses a SAR image classification method based on sparse representation and Gaussian distribution. The method comprises the following steps of (1) inputting a synthetic aperture radarSAR image; (2) generating the dictionary atomic matrix of a training set and a sparse coefficient matrix; (3) calculating the weight matrix of the sparse coefficient matrix and a binary matrix; (4) using the Gaussian distribution to generate the transition matrix of the dictionary atomic matrix; (5) updating the dictionary atomic matrix; (6) updating the weight matrix; (7) updating the binary matrix; (8) acquiring the sparse coefficient matrix; (9) determining whether a dictionary atomic matrix error reaches 10<-6>, if the dictionary atomic matrix error reaches 10<-6>, acquiring the dictionary atomic matrix of the trained training set and the trained sparse coefficient matrix and executing a step (10), otherwise, executing the step (4); (10) acquiring the classifier of the training set; and (11) classifying a test set. In the invention, a sparse representation and Gaussian modeling method is adopted and classification precision is increased.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a SAR (Synthetic Aperture Rader) image classification method based on sparse representation and Gaussian distribution in the technical field of image classification. The invention can be used to identify the target in the synthetic aperture radar SAR image, and can also be used to classify the ground objects in the synthetic aperture radar SAR image. Background technique [0002] Synthetic aperture radar SAR (Synthetic Aperture Radar) is a high-resolution imaging radar, which is widely used in military, agriculture, navigation, geographic surveillance and many other fields due to its unaffected by the environment, strong penetrating power, and high resolution. . Therefore, in the military field, synthetic aperture radar can detect important military targets such as armored vehicles, tanks, and aircraft. In the civilian field, it can also conduct research on the di...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that the classification dictionary matrix used is obtained by solving the sparse representation, and the error between the sample matrix and the dictionary is large after only one calculation, thereby reducing the classification accuracy of the synthetic aperture radar SAR image

The disadvantage of this method is that the simplified matching pursuit algorithm uses the linear operation of the atomic vector to gradually approach the signal vector, and the calculation of the atomic vector is a nonlinear problem. This method does not consider that the synthetic aperture radar SAR image obeys the complex distribution. , the relationship between pixels cannot be mined only through simple linear calculations, resulting in image information that cannot be fitted

Images