Mereduksi derau pada citra menggunakan teknik Neuro Fuzzy

HERLAMBANG, Sanam, Prof. Adhi Susanto, M.Sc.,Ph.D

2007 | Tesis | S2 Teknik Elektro

Dalam mengakuisisi citra seringkali citra mengalami degradasi (penurunan mutu), misalnya mengandung cacat atau derau (noise). Pemulihan/perbaikan citra berbasis Adaptive Neuro Fuzzy Inference System (ANFIS) atau teknik Neuro-fuzzy yang digunakan dalam penelitian ini, ditujukan untuk menghilangkan atau mereduksi derau pada citra berwarna dan citra keabuan dengan memanipulasi parameter premis dan konsekuen. Agar dapat lebih dimengerti oleh mesin atau manusia peningkatan nilai SNR harus cukup signifikan. Dari hasil proses simulasi diperoleh nilai-nilai galat kuadrat rata-rata (MSE) sebesar 67 - 277 atau kenaikan SNR sebesar 1,37 dB – 12,67 dB.

In many and practical image acquisition systems the resulting image data tend to be obscured by natural and equipment noises. An Adaptive Neuro-Fuzzy Inference System (ANFIS) was implemented in an effort to find its merit in reducing the noise in some color and gray-scale image, based on their premise and consequent parameters manipulation. For the purpose of further processing to expose the expected features of each image, significant increment in SNR was the goal of the experiment. The results show the residual (MSE) of 67 to 277 or SNR boosts of 1.37 dB to 12.67 dB were considered sufficient for the associated machine as well as human perceptions.

Kata Kunci : Pengolahan Citra,Teknik Neuro Fuzzy,Neuro-Fuzzy Systems (ANFIS), original image, noise, enhancement, SNR and (MSE)