Theorem 1: H-S transform in real domain (scalar) ex2=2πη∫exp(−2ηz2+2ηxz)dz,∀x,η proof: Given the Gaussian distribution N(x∣η2z,η1), we have following formula , thanks to the normalization of PDF, ∫2πηexp[−2η(x−η2z)2]dx=1 Expand the formula above, we get ∫2πηexp(−2ηx2+2ηxz)dx=ez2 Interchanging the x and z, we have finished the proof.
Theorem 2: H-S transform in real domain (scalar) e−x2=2πη∫exp(−2ηz2+j2ηxz)dz,∀x,η where j=−1.
proof: Similarly as the proof of theorem 1, we also use the normalization of Gaussian PDF. Given the Gaussian distribution as below N(x∣jη2z,η1)=2πηexp[−2η(x−jη2z)2] Using the normalization of PDF, we have ∫2πηexp[−2ηx2+j2ηxz]dx=e−z2 Interchanging x and z yields the theorem 2.
Theorem 3: e∥x∥2=∫(2πη)N/2exp(−2ηzTz+2ηxTz)dz proof: Given the vector Gaussian distribution N(x∣η2z,η1I)=(2πη)N/2exp[−2η∥x−η2z∥2] Using the normalization of Gaussian distribution, we have ∫(2πη)N/2exp[−2ηxTx+2ηxTz]dx=ezTz Interchanging the x and z yields theorem 3.
Theorem 4: e−∥x∥2=∫(2πη)N/2exp(−2ηzTz+j2ηxTz)dz proof: see the proof of theorem 2 and theorem 3.