基于形態學和分形理論的農產品圖像去噪

張宏群，陶興龍

摘要：結合形態學和分形理論對農產品圖像進行去噪處理，先通過形態學的開運算和閉運算濾除圖像的正脈沖和負脈沖噪聲，濾除噪聲的高頻部分效果明顯，再計算經過形態學濾波后圖像的分形維數，通過分形維數估算出修正參數修正圖像，不僅能兼并噪聲的低頻部分，還能去除由于形態學結構元素方向的限制所產生的毛刺，起到平滑圖像的作用。試驗仿真得到的結果與傳統的小波去噪、中值濾波、均值濾波相比較，不僅有較高的信噪比，還能保留圖像的細節，提高主觀的視覺效果。

關鍵詞：農產品圖像去噪；形態學；分形理論；信噪比

中圖分類號：TP391 文獻標識碼：A 文章編號：0439-8114（2013）05-1168-03

Agriculture Image Denoising Based on the Morphology and the Fractal Theory

ZHANG Hong-qun1，2，TAO Xing-long1

（1.Nanjing University of Information Science and Technology， Nanjing 210044， China；

2.Jiangsu Key Laboratory of Meteorological Observation and Information Processing ， Nanjing 210044， China）

Abstract： Combining the morphology and the fractal theory can effectively denoise the agriculture image. Using the opening and closing operations， the positive and negative pulse noises of the image can be wiped off. It indicates that the high frequency of noise denoised has an obvious effect. And then fractal dimension of the image processed by morphological filtering is obtained. Using correction parameter estimated by fractal dimension， the image should be corrected. It shows that the fractal theory can not only wipe off the low frequency part of noise， also play the role of smoothing image by removing the burr produced by direction restriction elements of morphology. At the same time， the experimental results compare to wavelet denoising， median filtering and mean filtering， it shows that the method not only have higher signal-to-noise ratio， also can reserve the detail of the image and improve the subjective visual effect.

Key words： agriculture image denoising； morphology； fractal theory； SNR