Research And Hardware Implementation Of FPGA-based Infrared Image Smoothing And Sharpen Algorithms

With the development of infrared imaging technology, higher requirements of real timeinfrared image processing should be met. For the good performances of hardware ataccelerating image processing, recently domestic or external, more and more infraredimaging systems choose Field Programmable Gate Array (FPGA) as their main imageprocessing units, and FPGA-based infrared image processing algorithms are researched.In China, most of the present work is to replant the traditional image processingalgorithm in FPGA, but less attention is paid on the research and improvement of thesealgorithms. In this area, we still have a lot of work to do in. For this, the main task of thispaper is to research and improve FPGA-based infrared image smoothing and sharpenalgorithms, and perform their hardware implementation.Usually, an image contains different noises, and single method can hardly treat them. Inexisting FPGA-based image smoothing algorithms, neighborhood averaging method canreduce gaussian noise effectively, while median filter is good at removing salt-pepperimpulse noise. This paper presents an adaptive smoothing method based on edge detection.On analysis of neighborhood characteristics, which chooses neighborhood averaging methodto process Gaussian noise and median filter to process salt-pepper noise respectively.Laplacian sharpening is an effective FPGA-base image sharpen algorithm. It couldbuildup image edge effectively, but because of its fixed sharpening template and condition,there may be excessive enhancement and edge distortion when processing some images. Inview of this situation, a configurable limited Laplacian sharpening algorithm is proposedwhich sharpening template and condition can be set to adapt to the edge enhancement ofdifferent image.Our improved FPGA-based infrared image smoothing and sharpen algorithms wereimplemented in FPGA EP3C40F780C6produced by Altera. These modules are high reliable,high-speed and portable. Compared with existing FPGA-base infrared image smoothing and sharpen algorithms, they can process image more flexibly and gain better results. Theirmaximum operation frequencies are more than150MHz, and this can satisfy the real timeimage processing requirement of1024×1280high resolution and100Hz high frame rateinfrared imaging system. They also have the advantage of low resource cost and simplestructure. They only occupy5%and4%resource of EP3C40F780C6respectively, whichmakes them easily embedded into FPGA-base infrared image processing system.