The effective image denoising method for MEMS based IR image arrays

摘要: MEMS have become viable systems to utilize for uncooled infrared imaging in recent years. They offer advantages due their simplicity, low cost and scalability high-resolution FPAs without prohibitive increase in cost. An thermal detector array with NETD is designed fabricated using bimaterial microcantilever structures that bend response change. The IR images of objects obtained by these FPAs are readout an optical method. For the images, processed a sparse representation-based image denoising inpainting algorithm, which generalizing K-Means clustering process, adapting dictionaries in order achieve signal representations. image quality improved obviously. Great compute analysis been realized discussed algorithm simulated data in applications on real data. experimental results demonstrate, better RMSE highest Peak Signal-to-Noise Ratio (PSNR) compared traditional methods can be obtained. At last we discuss factors determine the ultimate performance FPA. And indicated one unique advantages present approach is the larger arrays.