Consumer-level digital single-lens reflex (DSLR) cameras are typically not used in professional astronomy because of the systematic errors present in the data as a result of the strong intra- and interpixel variations associated with each of the three different colors (RGB) of the Bayer color filter array. Nevertheless, because the cost of DSLRs compared with traditional astronomical CCDs is so much lower, they represent a potentially underexplored area of scientific quality astronomical imaging, especially in the area of wide-field transient surveys. We demonstrate an algorithm that can achieve   ≈  1  %   level photometry in each of the RGB color channels from a stellar source and discuss the application of this algorithm to a ground-based transiting exoplanet survey. The algorithm primarily takes advantage of the large number of stellar sources available for statistical averaging within a single image, using a “lucky point-spread function” approach to identify sources in the image that exhibit systematic errors consistent with a chosen target from the same image. The selection of the appropriate“lucky” reference stars is accomplished through a comparison of the stellar image morphology as it appears on the Bayer array and the reference stars. These references are linearly combined to form a synthetic comparison star that can be used for differential photometry with the target. One key to the algorithm is that all data are retained at the individual pixel-level until the final differential comparison, which helps to alleviate systematic effects that might otherwise cancel each other out during the flux-summing process. We demonstrate the algorithm on HD 339461, a m_V  =  8.93 G0-type star on which we achieve single-percent level photometry that approaches the fundamental noise floor possible from a single camera.