曾禎宇
Separated vs. Holistic R G B Color Image
Separated vs. Holistic RGB Color Image
Reference T. A. Ell and S. J. Sangwine, “Hypercomplex Fourier transforms of color images,” IEEE Transactions on Image Processing, vol. 16, pp. 22–35, Jan T. A. Ell, “Multi-vector color-image filters,” in Proceedings of the IEEE International Conference on Image Processing, vol. V, (San Antonio, Texas),pp. 245– 248, Sept Geometric Algebra: se99/ se99/
Outline Geometric Algebra Multi-Vector Color Images Applications
Quaternion based Color Image Representation For 3-D Space: 4 Elements {1, i, j, k} For Color Image f(n,m) = r(n,m) i + g(n,m) j + b(n,m) k Hypercomplex Representation i j k
Geometric Algebra for Color Image Representation Quaternion: f(n,m) = r(n,m) i + g(n,m) j + b(n,m) k Geometric Algebra f(n,m) = r(n,m) e 1 + g(n,m) e 2 + b(n,m) e 1 ^ e 2 More General Quaternion is the special case of Geometric Algebra
Geometric Algebra Inner Product a · b = |a||b| cos θ Outer Product a ^ b : bi-vector magnitude of a ^ b = |a||b| sin θ a ^ b = - b ^ a a^(b+c) = a^b + a^c Geometric Product ab = a · b + a^b ba = a · b - a^b
Geometric Algebra in 2-D 2 orthonormal vectors e 1 and e 2 e 1 2 = e 2 2 = 1 e 1 · e 2 = 0 A full algebra is spanned by 1 scalar1 2 vectors {e 1, e 2 } 1 bi-vector e 1 ^ e 2
Multi-Vector A = a 0 + a 1 e 1 + a 2 e 2 + a 12 e 1 ^e 2 e 1 ^e 2 = I 2 (pseudo-scalar) I 2 2 = e 1 e 2 e 1 e 2 = -e 2 e 1 e 1 e 2 = -1 A = a 0 + a 1 e 1 + a 2 e 2 + a 3 I 2 A = (a 0 + a 12 I 2 ) + ( a 1 e 1 + a 2 e 2 ) (a 0 + a 12 I 2 ) : a complex number
Multi-Vector for Color Image Representation A = (a 0 + a 12 I 2 ) + ( a 1 e 1 + a 2 e 2 ) f = L + v
Color Edge Detection Filters Example: 3 Type Horizontal Edge Detection
Color Edge Detection Filters Example: 3 Type Horizontal Edge Detection
Color Edge Detection Filters H1 detects edge in both luminance and chrominance H2 detects edge in chrominance but smoothes luminance H3 detects edge in luminance but smoothes chrominance