Generation of single image stereograms based on stochastic textures

Single image stereograms (SIS) create a vivid illusion of depth from a 2D image based on the stereo vision process. The method does not require special equipment, and has attracted attention in recent years. Here, the principles of SIS are discussed, as well as the most well known generating methods. We introduce a modification to the random-dot based algorithms, that leads to SIS patterns with less high-frequency content. This improves the 3D capture effect and the aesthetic aspect of the SIS, and avoids ink-spreading problems in high resolution output devices. A correlation function that measures the local match between the images seen by each eye is computed. This allows a direct evaluation of the correctness of the algorithms.