The comparison ofrecently suggested algorithms of fast approximate Karhunen-Loeve(KL) transform when applied to pattern recognition problem is presented. It is known that adequate signs selection is stiff the problem. In the paper we consider several algorithms that can perform fast and quafltative signs selection. Among them are wavelet based algorithm of approximate XL transform, 2D algorithm ofKL transform. algorithm with projection into proper orthogoxialized basis, and real tinie algorithm ofKL transform. To compare quality ofthe algorithms they are apphed to human facts tecognition prob1m. Th evem'y of the all abovementioned algorithms is demonstrated.