In this study, the second harmonic generation (SHG) response to polarization and subsequent data analysis is used to discriminate, in the same image, different SHG source architectures with pixel resolution. This is demonstrated in a mammalian tissue containing both skeletal muscle and fibrilar collagen. The SHG intensity variation with the input polarization (PSHG) is fitted pixel by pixel in the image using an algorithm based on a generalized biophysical model. The analysis provides the effective orientation, theta(e), of the different SHG active structures (harmonophores) at every pixel. This results in a new image in which collagen and muscle are clearly differentiated. In order to quantify the SHG response, the distribution of theta(e) for every harmonophore is obtained. We found that for collagen, the distribution was centered at theta(e) = 42.7 degrees with a full width at half maximum of theta = 5.9 degrees while for muscle theta(e) = 65.3 degrees , with theta = 7.7 degrees . By comparing these distributions, a quantitative measurement of the discrimination procedure is provided.