第五章 實驗結果
31
能用小波轉換分解的最佳階數,並從解析出來的頻帶中尋找出最佳的 特徵向量(feature vector),以期在不影響辨識率的情況下,利用最 少的特徵向量來進行指紋辨識,達到有效降低計算量並藉此增快比對 速度的效果,實驗模擬結果如表一所示。
從實驗的結果來進行討論分析,發現指紋影像在空間域的位移偏 差並不會對頻率域特徵值的辨識造成太大的影響,確認影像從空間域 經過離散小波轉換到頻率域之後,用統計學的方法擷取特徵值的方法 是可行的。實驗結果顯示雖然一階分解的辨識率並不高,但隨著分解 階數的增加,辨識率明顯的提升,其中,以第三階分解的辨識效果最 佳,第四階次之。結果顯示藉由多解析(multi-resolution)的方式可 提高辨識率,但並不是越多層就越好,在此以小波轉換為基礎的指紋 辨識系統中,第三階分解有最佳辨識率。
實驗結果發現利用 HL 和 LH 兩個頻帶的辨識率,與利用 HL、LH 和 HH 三個頻帶的辨識率幾乎相同,因此為了加快辨識的處理速度,
減少計算量,可以忽略 HH 頻帶,只擷取 HL 和 LH 頻帶的特徵值進行 辨識,如此一來則可以在不影響辨識率的情況下,對特徵向量進行最 佳化,減少不必要的運算達到增進辨識處理速度的目的。
第五章 實驗結果
32
表二所示為以小波轉換為基礎的指紋辨識系統、circular Gabor decomposition 與 block Gabor decomposition 等三種指紋辨識系統 進行比較的結果。從實驗結果得知,利用小波轉換進行分解所擷取的 特徵向量長度比使用 Gabor filters 所得到的特徵向量較短,相當有 利於增進辨識的反應時間。此外,以小波轉換所得到的特徵向量可藉 由尋找最佳特徵向量的方法,在不影響辨識率的條件下,調整特徵向 量長度(如: 忽略 HH 頻帶的特徵值)。綜合實驗結果得知,以小波轉 換為基礎的指紋辨識系統有良好的辨識率,且能大幅降低特徵向量長 度,減少記憶體的使用量,快速的完成指紋辨識工作。
第五章 實驗結果
33
Figure 5.1 指紋影像一階分解
Figure 5.2 指紋影像四階分解
第五章 實驗結果
34
96%
98%
93%
48%
Recognition rate
4
thlevel 3
rdlevel 2
ndLevel 1
stlevel
96%
98%
93%
48%
Recognition rate
4
thlevel 3
rdlevel 2
ndLevel 1
stlevel 各階辨識率
表一: 小波轉換各階分解辨識率
90%
complex fix long Gabor filters (circular)
84%
98%
Recognition rate
middle simple
Computation
fix adjustable
Vector length
middle small
Feature vector
Gabor filters (block) Wavelet
transform
90%
complex fix long Gabor filters (circular)
84%
98%
Recognition rate
middle simple
Computation
fix adjustable
Vector length
middle small
Feature vector
Gabor filters (block) Wavelet
transform
Wavelet transform compare with Gabor filters
表二: 三種指紋辨識法的比較
第五章 實驗結果
35
5-2 討論與分析
小波轉換是對原始影像資訊經過擴張/收縮及平移頻域分離處理 的方式進行轉換處理,它能有效的將輸入資訊的低頻與高頻資料做分 離處理,在解析度方面,低頻資訊可繼續做小波轉換分離出下一層的 低頻與高頻資訊,隨著分解越多,則可得到越多的子頻帶,這些不同 頻域的小波係數,代表著頻域中各自範圍資訊,這種分離方式可對資 訊進行有效的管理與處理,用於指紋辨識上,有相當好的效果。
第六章 結論
36
第六章 結 論
利用離散小波轉換來進行指紋辨識,能快速且有效的分解出指紋 影像的各個子頻帶資訊,迅速的完成指紋辨識工作。與特徵點辨識法 相比可節省更多的前處理(ex: image enhancement, directional filtering, ridge segmentation, ridge thinning)。與使用 Gabor filters 的指紋辨識法相比,擁有較小的特徵向量,所需的計算量較 低。本論文所設計的小波轉換指紋辨識系統可以有效地減少特徵值數 量,以較短的特徵向量降低計算複雜度以加快辨識的速度。
參 考 文 獻
37
參 考 文 獻
[1] “Wavelet domain features for fingerprint recognition”
Tico, M.; Kuosmanen, P.; Saarinen, J.;
Electronics Letters , Volume: 37 , Issue: 1 , 4 Jan. 2001 Pages:21 - 22
[2] “Fingerprint recognition using wavelet features”Tico, M.;
Immonen, E.; Ramo, P.; Kuosmanen, P.; Saarinen, J.;
Circuits and Systems, 2001. ISCAS 2001. The 2001 IEEE International Symposium on , Volume: 2 , 6-9 May 2001 Pages:21 - 24 vol. 2
[3] “A wavelet-based method of nearest neighbor pattern
classification using scale sequential matching”Creusere, C.D.;
Hewer, G.;Signals, Systems and Computers, 1994. 1994 Conference Record of the Twenty-Eighth Asilomar Conference on , Volume: 2 , 31 Oct.-2 Nov. 1994
Pages:1123 - 1127 vol.2
[4] “Biometric personal identification based on iris patterns”
Yong Zhu; Tieniu Tan; Yunhong Wang;
Pattern Recognition, 2000. Proceedings. 15th International Conference on , Volume: 2 , 3-7 Sept 2000
Pages:801 - 804 vol.2
[5] “Image coding using wavelet transform” Antonini, M.; Barlaud, M.; Mathieu, P.; Daubechies, I.; Image Processing, IEEE
Transactions on , Volume: 1 , Issue: 2 , April 1992 Pages:205 – 220
[6] “Image coding using wavelet transform and classified vector quantization”Lin, S.; Salari, E.; Vision, Image and Signal Processing, IEE Proceedings- , Volume: 143 , Issue: 5 , Oct.
1996
Pages:285 – 291
參 考 文 獻
38
[7] “Texture feature for image classification” R.m.Haralick, K.Shanmugam and I.Dinstein, IEEE transactions on System, Man, Cybernatics, Vol.8, No.6,1973,pp.610-621.
[8] “Wavelet-based statistical signal processing using hidden Markov models”Crouse, M.S.; Nowak, R.D.; Baraniuk, R.G.;
Signal Processing, IEEE Transactions on [see also Acoustics, Speech, and Signal Processing, IEEE Transactions on] , Volume:
46 , Issue: 4 , April 1998 Pages:886 - 902
[9] “Statistical analysis of image differences by wavelet decomposition”Ruttimann, U.E.; Unser, M.; Rio, D.;
Engineering in Medicine and Biology Society, 1994. Engineering Advances: New Opportunities for Biomedical Engineers.
Proceedings of the 16th Annual International Conference of the IEEE , 3-6 Nov. 1994 Pages:A28 - A29 vol.1
[10] “Wavelets, statistics, and biomedical applications” Unser, M.;
Statistical Signal and Array Processing, 1996. Proceedings., 8th IEEE Signal Processing Workshop on (Cat.
No.96TB10004 , 24-26 June 1996 Pages:244 - 249
[11] “Hierarchical Markov models for wavelet-domain statistics”
Azimifar, Z.; Fieguth, P.; Jernigan, E.;Statistical Signal Processing, 2003 IEEE Workshop on , 28 Sept.-1 Oct. 2003 Pages:258 - 261
[12] “Towards random field modeling of wavelet statistics”
Azimifar, Z.; Fieguth, P.; Jernigan, E.;
Image Processing. 2002. Proceedings. 2002 International Conference on , Volume: 1 , 22-25 Sept. 2002
Pages:I-361 - I-364 vol.1
[13] “Texture synthesis via matching first and second order statistics of a wavelet frame decomposition”van Nevel, A.; Image
Processing, 1998. ICIP 98. Proceedings. 1998 International Conference on , Volume: 1 , 4-7 Oct. 1998 Pages:72 - 76 vol.1
參 考 文 獻
39
[14] “Analysis of functional magnetic resonance images by wavelet decomposition”Ruttimann, U.E.; Ramsey, N.F.; Hommer, D.W.;
Thevenaz, P.; Chulhee Lee; Unser, M.;Image Processing, 1995.
Proceedings., International Conference on , Volume: 1 , 23-26 Oct. 1995 Pages:633 - 636 vol.1
[15] “Generalized wavelet transforms and the cortex
transform”Martin, R.; Cochran, D.; Signals, Systems and Computers, 1994. 1994 Conference Record of the
Twenty-Eighth Asilomar Conference on, Volume: 2 , 31 Oct.-2 Nov. 1994 Pages:1105 - 1108 vol.2
[16] “IEE Colloquium on `Applications of Wavelet Transforms in Image Processing' (Digest No.009)”Applications of Wavelet Transforms in Image Processing, IEE Colloquium on , 20 Jan 1993
[17] “A little introduction to wavelets” Nason, G.P. ;Applied Statistical Pattern Recognition (Ref. No. 1999/063), IEE Colloquium on , 20 April 1999
Pages:1/1 - 1/6
[18] “Discrete frequency warped wavelets: theory and applications”
Evangelista, G.; Cavaliere, S.; Signal Processing, IEEE Transactions on [see also Acoustics, Speech, and Signal
Processing, IEEE Transactions on] , Volume: 46 , Issue: 4 , April 1998 Pages:874 – 885
[19] “Wavelet function, scaling function and discrete wavelet
transform”Jin, Q.; Wong, K.M.; Luo, Z.Q.;Time-Frequency and Time-Scale Analysis, 1994., Proceedings of the IEEE-SP
International Symposium on , 25-28 Oct. 1994 Pages:409 – 412
[20] “Fingerprint enhancement”Lin Hong; Jian, A.; Pankanti, S.;
Bolle, R.;Applications of Computer Vision, 1996. WACV '96., Proceedings 3rd IEEE Workshop on , 2-4 Dec. 1996
Pages:202 – 207