ファジィ推論を利用した画像の平滑化 ＩＩ 応用 [in Japanese] Image Smoothing Based on Fuzzy Reasoning. II. Application. [in Japanese]
Access this Article
画像の平滑化において相反する要素である統計的雑音除去とエッジ保存の双方を満足させるため, ファジィ推論を利用した平滑化の方法を提案し, これを数値ファントムならびに実際の核医学画像に応用した。<BR>この方法を繰り返し適用してもエッジが保持され, 同時に雑音も除去される平滑化画像が得られた。これは, 本平滑化アルゴリズムの有用性を示すものである。また, その計算時間は実用的応用上の妥当な範囲内にあり, 本方法の高い実用性を示唆する。
<I>A smoothing method based on fuzzy reasoning has been proposed to well combine two conflicting factors in image smoothing, noise removal and edge preservation. This smoothing algorithm is described by three linguistic rules called fuzzy reasoning rules, thus differing from usual smoothing algorithms based on analytical rules. From the three reasoning rules used, a proper way of smoothing for each pixel was inferred when the magnitude of gradient at the pixel was given</I>.<BR><I>To evaluate its effectiveness, the fuzzy smoothing algorithm was applied to a simulated digital phantom image and real nuclear medicine images. Smoothing algorithm is evaluated from both its noise removal ability and edge preservation ability. How well smoothing algorithm can combine the two abilities was examined by increasing the number of smoothing iterations. Furthermore, the computation time of the fuzzy smoothing algorithm is measured to evaluate its practicality. The results are compared with those obtained by three smoothing algorithms described by analytical rules, such as a weighted averaging and a median filtering. These two have widely been used for smoothing for nuclear medicine images. All algorithms including the fuzzy smoothing one are performed on a 3×3 window</I>.<BR><I>From smoothed results obtained by the smoothing algorithms, it is found that the fuzzy smoothing algorithm achieved the best combination of noise removal and edge preservation, and its computation time was reasonable. It is thus concluded that the fuzzy smoothing algorithm is useful and has a high potential for practical applications</I>.
RADIOISOTOPES 41(10), 491-498, 1992
Japan Radioisotope Association