Dual tree fractional quaternion wavelet transform for disparity estimation

详细信息    查看全文

• 作者：Sanoj Kumar^a ; ^{sanoj4ever@gmail.com" class="auth_mail} ; ^{sanojdma@iitr.ernet.in" class="auth_mail} ; Sanjeev Kumar^a ; ^{malikfma@iitr.ernet.in" class="auth_mail} ; Nagarajan Sukavanam^a ; ^{nsukvfma@iitr.ernet.in" class="auth_mail} ; Balasubramanian Raman^b ; ^{balarfma@iitr.ernet.in" class="auth_mail}
• 关键词：Fractional quaternion wavelet transform ; Complex wavelet transform ; Fractional Hilbert transform ; Quaternion algebra ; Disparity estimation ; Optical flow
• 刊名：ISA Transactions
• 出版年：March, 2014
• 年：2014
• 卷：53
• 期：2
• 页码：547-559
• 全文大小：3550 K

文摘

This paper proposes a novel phase based approach for computing disparity as the optical flow from the given pair of consecutive images. A new dual tree fractional quaternion wavelet transform (FrQWT) is proposed by defining the 2D Fourier spectrum upto a single quadrant. In the proposed FrQWT, each quaternion wavelet consists of a real part (a real DWT wavelet) and three imaginary parts that are organized according to the quaternion algebra. First two FrQWT phases encode the shifts of image features in the absolute horizontal and vertical coordinate system, while the third phase has the texture information. The FrQWT allowed a multi-scale framework for calculating and adjusting local disparities and executing phase unwrapping from coarse to fine scales with linear computational efficiency.