Research On Algorithm Of Electromagnetic Interference Detection With FY-3 MERSI Data

Electromagnetic interference including abnormal data block,random value impulse noise and stripe noise,is widespread in FY-3 MERSI data under the significant influence of the complex space electromagnetic environment and the instrument internal noise.It seriously disturbs application and development of scientific research and social produce.Hence,electromagnetic interference detection is necessary and considerable.In this paper,the work research on electromagnetic interference detection in FY-3 MERSI data is carried out,mainly including the following contents:(1)We analyze the generation mechanism and distribution characteristics of electromagnetic interference,and summarize that there are three main kinds image noise in FY-3 MERSI data including abnormal data block,random value impulse noise and stripe noise.Mathematical models are established to provide a theoretical basis for electromagnetic interference detection.(2)Since a clear line of demarcation between abnormal data block and normal data,the horizontal or vertical gradient expectation is assessed.We use the threshold method and edge detection method detect the abnormal data block.The experimental results show that the threshold method has better detection performance than the edge detection method for large-span abnormal data blocks,while the edge detection method can be used as a supplementary algorithm for small-span abnormal data blocks.(3)A random value impulse noise rate evaluation method based on checkerboard segmentation and low roughness region selection is proposed to evaluate the noise rate of FY-3 MERSI data to provide theoretical support for threshold selection.We study the typical random value impulse noise detection mechanisms such as the median filter structure,absolute difference ordering,and detail image structure window filtering,and evaluate the advantages and disadvantages of the detection mechanism.The experimental results show that the detailed image structure window filter can obtain the best random value impulse noise detection result,and objectively and accurately protect the edge structure and texture information of the FY-3 MERSI image.(4)We study typical stripe noise detection mechanisms such as histogram matching method,moment matching method and its improvement method,Fourier transform method and wavelet transform method,aiming at resolution data of periodic bands in the space domain and the characteristics of the frequency domain.The experimental comparative analysis of electromagnetic interference detection and the existing problems,the results show that the algorithm based on the mean value of the detection of band noise is better than other algorithms.This paper selects the optimal algorithm to detect the abnormal data block,random value impulse noise and stripe noise in the FY-3 MERSI data,and achieves the purpose of detecting electromagnetic interference.It improves the accuracy of the inversion product of the FY-3 MERSI data.