We propose a soft sphere detection algorithm where search-bounds are determined based on the distribution of candidates found inside the sphere for different search levels. Detection accuracy of unbounded search is preserved ...