The image denoising method for MEMS based uncooled infrared imaging system

The uncooled infrared imaging based on MEMS has more and more broad space for development in recent years. An uncooled thermal detector array was designed and set up using bi-material micro-cantilever structures, which can bend with the temperature change. The effective image points of objects' infrared images which are read out by an optical method from this thermal detector array are discrete. For this reason, the output image should be filtered based on the gray mean value of square window, firstly. Then, each point of image can be decided to assign zero or restore the initial gray, according to the threshold of gray value summation of the filtered image's single direction template. Comprehensive two directions' data that is in horizontal and vertical, the final result is achieved. The experimental results demonstrate, this algorithm can remove noise well without losing the details of objects' effective image points.