Polarized SAR image classification method based on deep complex value full convolutional neural network

• Main Authors: LIANG WENKAI, ZHANG PENG, LI MING, CAO YICE, WU YAN
• Format: Patent
• Language: Chinese; English
• Published: 24.01.2020
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a polarized SAR image classification method based on a depth complex value full convolutional neural network, and mainly solves the problem of low classification precision in the prior art. According to the scheme, the method comprises the steps of inputting a polarization coherence matrix T and a real ground object mark G of a to-be-classified polarized SAR image, and normalizing the T; extracting an input complex value vector of the normalized matrix to construct a feature matrix F; selecting pixel points in the G to generate a new real ground object mark G'; respectively generating a feature set and a mark set on the F and the G' through a sliding window, and randomly selecting the feature set and the mark set to form a training set; constructing a deep complex value full convolutional neural network, and initializing the deep complex value full convolutional neural network; training the initialized deep complex value full convolutional network by using the training set; inputting t