A validation of the cosmic distance duality (CDD) relation, η(z) ≡ (1+z) 2 d A (z)/d L (z) = 1, coupling the luminosity (d L ) and angular-diameter (d A ) distances, is crucial because its violation would require exotic new physics. We present a model-independent test of the CDD, based on strong lensing and a reconstruction of the HII galaxy Hubble diagram using Gaussian Processes, to confirm the validity of the CDD at a very high level of confidence. Using parameterizations η(z) = 1 + η 0 z and η(z) = 1 + η 1 z + η 2 z 2 , our best-fit results are η 0 = 0.0147 +0.056 −0.066 , and η 1 = 0.1091 +0.1680 −0.1568 and η 2 = −0.0603 +0.0999 −0.0988 , respectively. In spite of these strong constraints, however, we also point out that the analysis of strong lensing using a simplified single isothermal sphere (SIS) model for the lens produces some irreducible scatter in the inferred CDD data. The use of an extended SIS approximation, with a power-law density structure, yields very similar results, but does not lessen the scatter due to its larger number of free parameters, which weakens the best-fit constraints. Future work with these strong lenses should therefore be based on more detailed ray-tracing calculations to determine the mass distribution more precisely.