Performance Analysis of FPGA Based Modified Adaptive Median Filter (M-AMF) Design for Medical Applications
Archana H. R, C R Byra Reddy, CP Narendra · 2019
The filtering mechanism in digital image processing is utilized to remove the undesirable elements or features from the image. The median filtering is a popular filter technique to suppress the undesirable impulse noises. In this paper, an efficient Modified- Adaptive Median Filter (M-AMF) is designed to filter the impulse salt-and-pepper (SP) noise at different levels 1%, 5 % and 10% for medical applications. The modified-AMF method using three-input sorter includes window 3×3 module, median filtering, and adaptive computation with error detection module. The median filter uses the sorter module to find the median value. The adaptive computation module obtains either the median filtered output or center pixel value from the window module, based on the adaptive condition. Error detection module obtains the corrupted pixels from the filtered output image. To evaluate the performance metrics of Modified-AMF on Xilinx platform, PSNR and MSE are calculated for medical Brain MRI image with the different noise level and also a comparison of similar exiting methods with an improvement of hardware constraints includes area and maximum operating frequency is tabulated. The proposed M-AMF model is synthesized on Xilinx ISE 14.7 and simulated on ModelSim and Implemented on Artix-7 FPGA.