The monograph outlines a method of distinguishing people in low-resolution television images using a histogram oriented gradients, a nonlinear processing model consisting of an adaptive interpolating and retrainable twelve- to twenty-seven dimensional artificial neural network to help reduce losses in television images, a mobile device that narrows the transmission band by compacting real-time television images by size and a hardware-software complex has been developed based on an improved method of identifying people in television images used for search purposes in emergencies.