Fuzzy SVM based pre-processing technique for infrared (IR) thermal images

摘要

This paper proposes a noise removal algorithm to improve the quality of images captured by IR thermal cameras. Due to atmospheric disturbances and the temperature reflected from the surrounding objects, IR images are generally noisy. For removal of noise in IR thermal images the proposed algorithm has two stages. In the first stage, a support vector machine incorporated with fuzzy logic (FSVM) is used to obtain noise pixels from the thermal image and in the second stage an adaptive directional weighted median filter (ADWMF) is applied to restore noisy pixels. Our proposed algorithm is compared with another fuzzy-logic based noise classification technique, namely the adaptive fuzzy inference based directional median filter (AFIDMF). Simulation results indicate that the performance of the proposed filter is better, compared to AFIDMF based on the measures such as peak signal to noise ratio (PSNR) and structural similarity index (SSIM).