We present an application of the Richardson-Lucy algorithm to the analysis of color-magnitude diagrams by converting the CMD into an image and using a restoring point spread function function ({\it psf}) derived from the known, often complex, sources of error. We show numerical experiments that demonstrate good recovery of the original image and establish convergence rates for ideal cases with single gaussian uncertainties and poisson noise using a $χ^2$ statistic. About 30-50 iterations suffice. As an application, we show the results for a particular case, the Hipparcos sample of the solar neighborhood where the uncertainties are mainly due to parallax which we model with a composite weighted gaussian using the observed error distributions. The resulting psf has a slightly narrower core and broader wings than a single gaussian. The reddening and photometric errors are considerably reduced by restricting the sample to within 80 pc and to M_V ≤ 3.5. We find that the recovered image, which has a narrower, better defined main sequence and a more clearly defined red giant clump, can be used as input to stellar evolution modeling of the star formation rate in the solar vicinity and, with more contributing uncertainties taken into account, for general Galactic and extragalactic structure and population studies. Comment: Accepted for publication in Astronomy and Astrophysics, main journal