碎形方影應用於影像壓縮與分析上之研究

Abstract

數值影像處理不但使人們對影像資訊更容易了解, 也使得影像資料 可以隨著自動化機器(如電腦)的發展而有更廣範的應用. 然而, 不論是在 數值影像的傳輸或是儲存方面, 均會耗去上述自動化機器相當大的資源( 如電腦的硬碟空間), 於是, 資料壓縮技術在數值影像處理中便愈發顯得 重要. 本研究探討碎形方法於影像壓縮上之應用. 研究中, 使用碎形 壓縮程式及JPEG對兩幅SPOT-2全色態影像及一幅航照影像分別作壓縮,並 針對各壓縮影像計算其與原始影像間之關聯係數, 均方根誤差, 像元灰度 值差異, 訊號雜訊比及熵值等以比較其統計特性. 另外, 利用碎形壓縮過 程中區塊的分類辨識出影像的特徵. 最後, 對不同壓縮程式之壓縮時間作 評估. 成果顯示, 碎形壓縮往往可以得到較大之壓縮比例, 而當壓縮 比例達到一定程度時, 碎形壓縮影像之品值將會優於JPEG. 而在特徵分析 方面, 影像中灰值相近的大區域較易分辨出其邊界. 壓縮時間方面, 碎形 壓縮則明顯偏高. The image processing technique provides more acceptable informations.On the computer's development, the technique applied on image data becomesmore and more commanly. However, when digital image data is transmittingor stroing, it would spend much more resource, e.q. hard disk capacity. For this reason, data compression technique is more important on the digi-tal image processing field. The thesis discusses image compression which is based on fractal method,and image features are analyzed when fractal compression is proceeding. The performance of two fractal based image compression programs are compara-tively studied with JPEG implementation. Numerical experiments are conductedwith two SPOT panchromatic sub scenes and one photograph. Statistic indices,including the entropy, correlation coefficient, root mean square error, pixel value differences, and signal noise ratio, are computed from differences be-tween the compressed and the original images. The results indicate that fractal compression methods provide higher com-pression ratio. In general, when the compression is higher than 50, fractalbased approaches are better than JPEG. The result of feature analysis showsthat edges of homogeneous and large polygon data will easier to reognize.