FPGA Based SAR Flood Image Change Detection

Change detection is widely used in environmental detection,forest monitoring,military monitoring,disaster assessment and other fields.It is an important application direction of remote sensing imagery.In order to meet the real-time and accuracy requirements of flood detection applications,and to realize SAR image flood monitoring on board,this paper proposes a high-precision SAR flood image change detection algorithm suitable for FPGA hardware acceleration and completes FPGA hardware Design and simulation,the main work is as follows:In order to avoid the shortcomings of the existing wavelet-based detection algorithm,such as high complexity,low automation and difficult to implement onboard,a change detection method based on Log-Ratio Wavelet Multi Scale Fusion is proposed.The method uses edge detection instead of complex local coefficient of variation to solve the complexity problem,and uses the difference image statistical histogram to select a reasonable threshold to complete the automatic threshold segmentation.The experimental results show that the proposed scheme maintains the accuracy of the existing wavelet scheme and greatly reduces the complexity of the algorithm.and automatic change detection are implemented.Based on the improved wavelet multi-scale analysis method,this paper proposes a change detection method based on the Fusion of Likelihood Ratio and Log-Ratio difference image to avoid the defect of single pixel difference map.On the basis of LogRatio difference image,the regional information of the neighboring pixel points is integrated,which improves the background noise immunity.The experimental results show that the scheme can further suppress the background noise and retain enough good image details and higher detection accuracy.Based on the two improved change detection algorithms,the FPGA hardware implementation and Vivado simulation of the algorithm are carried out respectively,and the simulation results of FPGA are compared with the results of MATLAB algorithm to verify the effectiveness of FPGA implementation.The simulation results show that the wavelet-based method: FPGA and PC processing results are basically the same,the classification difference pixel is less than 20;the processing speed of FPGA is 82 times that of PC;and the most consumed hardware resource on XC7VH580 T chip is Memory LUT,which is 74.44%.When it comes to the fusion method,the FPGA and PC processing results are basically the same,the classification difference pixel is less than 8;the processing speed of the FPGA is 98 times that of the PC;and the most consumed hardware resource on the XC7VH580 T chip is the Memory LUT,which is 83.06%.