Cognitive radio (CR) is one of the most important challenges in contemporary communication. Because cognitive technologies are based on matching transmission parameters to the surrounding radio environment, several practical problems related to weak signal detection must be solved. These include, e.g., uncertainty of the noise variance value, the changes of this value in time and space, limited hardware resources or time of detection. All of these factors can seriously degrade system performance. The search for practically useful solutions of the encountered difficulties is therefore very important for the successful development and implementation of CR technology. This article provides an overview and comparison of chosen blind detection methods proposed for use in CR receivers. The comparison is based on simulation results, and the methods are compared in terms of detection efficiency, time complexity and robustness to changes in transmission parameters.