Point object recognition — Some single- and multi-channel applications

High-precision photometric analysis of images of crowded stellar fields needs sophisticated algorithms. The photometric precision is, however, a strong function of the completeness of source detection. We discuss several aspects of this problem, both with relation to single- and multi-channel applications. In single images, we separate detection into two phases, source image enhancement and actual detection. Comparative tests show a point spread function pixel fitting technique to give the best results. For the fraction of undetectable stars still affecting image statistics, we have developed a technique to extract information about their effect on the faint-end luminosity function. We give two examples of multi-channel image fusion applications: (1) a combination of low- and high-resolution images and (2) removal of undersampling effects by sub-pixel image displacements. Preliminary results show considerable potential for these techniques.