caiya55 / jitter-compensation

Remote Sensing Image Jitter Detection and Compensation Using Convolutional Neural Network

jitter-compensation

Remote Sensing Image Jitter Detection and Compensation Using Convolutional Neural Network

This the code from the paper Remote Sensing Image Jitter Detection and Compensation Using Convolutional Neural Network

We use neural networks to correct the deformed remote sensing images.

Framework: python3.5 + tensorflow1.4 + keras the most latest version

Platform: Windows Spyder

Proposed architecture:

Figure 1. Architecture of the proposed framework

CNN framework

Figure 2. Architecture of the proposed neural network

Experiment results on real remote sensing images are expressed below:

Figure 3. Left: Raw deformed image; Middle: Resotred image by proposed method. Right: Resoted image by on board sensors

Figure 4. Left: Raw deformed image; Middle: Resotred image by proposed method. Right: Resoted image by on board sensors

Figure 4. Left: Raw deformed image; Middle: Resotred image by proposed method. Right: Resoted image by on board sensors