A modified Canny Edge Detection Based on the Significance of Multiresolution

[[abstract]]邊緣是影像灰度劇烈變化的區域，也是資訊之所在。所有的特徵皆是集合眾多邊緣資訊所構成；如樣式(Texture)和圖形(Pattern and Shape)。邊緣的偵測與強化是影像處理的一主要目的。 邊緣與雜訊皆具高頻成分。邊緣偵測很容易受到雜訊的干擾，解決之道一般採提高信號雜訊比(S/N)策略，Canny法則採雙門檻值(Threshold value)即一典型的方式。然而Canny的方法著重細節的描述，對圖形辨識的應用常製造過多的累贅資訊。基於有意義的特徵，通常具有由高解析影像傳遞至低解析影像的特性，本文結合三階小波分解與Canny細線化策略，以有效達到特徵資訊壓縮的目的。三階小波分解提供重要邊緣特徵的定位，Canny細線化策略則用於尋找精確邊緣位置。實驗證明，此法可有效保留重要邊緣資訊，並減少55.45%的不必要資訊。 (關鍵詞 ：影像處理、小波邊緣偵測 影像處理、小波邊緣偵測 影像處理、小波邊緣偵測 影像處理、小波邊緣偵測 影像處理、小波邊緣偵測 )[[abstract]]Edge is theregion on whichthe information liesand theimagegrayscale changes dramatically. Anyfeature iscomposed of many edge informationsuch as texture, pattern and shapecollectively. The edge detection and enhancementis a major objective of image processing. Both edge and noise has the feature of the high frequency component,the edge detection issusceptible to the noise interference. generally the solution of the problem, adopted is to improve the signal to noise ratio (S / N) strategy.Canny rules adoptsdual threshold value that is a typical way. However “Canny method” focuses on the descriptionof the details,while the applicationof the pattern recognition often makestoo much baggage information. Based on the meaningful feature usually characterized by the transmission ofa high resolution imageto a lowresolution image,it is arrangedto combine the thirdorder wavelet decomposition with the Canny thinning strategy in order to effectively achieve the purpose of the compression of the characteristic information for this thesis.The thirdorder wavelet decompositionprovide important feature positioning.Canny’s thinning strategy is used to find the exact position of the edge. III Experiments show that this method can effectively preserve important edgeinformation, and reduce 55.45% of unnecessary information. (Keyword：Image Processing, Wavelet, Edge Detection