Quaternion Neural Network Simulator

The Quaternion Nerula Network Simulator (QNN Simulator) is developed based on Matlab.

If you want to try QNN simulator, please contact to us

email: shangfang@uec.ac.jp

The main features of the QNN Simulor are as follows.

★　If you need Poincare Sphere parameter (PSP) as the basic parameter for trainning and testing, you can use the PSP Supporter which is included in the Simulator to calculate them from original SAR data or other fully polarimetric radar data. Of course, you can also make your quaterion parameters for use as you like.

Fig.1 PSP Supporter for calculating Poincare Sphere Parameters

★　You can extract samples for trainning in the Simultor by giving coordinates, or you can directly use your prepared　samples.

Fig.2 Sample preparing panel of QNN simulator

★　In the training process, you can set up parameters of neural network such as neuron number, activation function, etc.

Fig.3 Training process panel of QNN simulator

★　In the validation process, you can set up the decision logic and RGB code for generating final image.

Fig.4 Validation process panel of QNN simulator

PAGE TOP