Detekce obličejů pomocí příznakového rozpoznání

Abstract

The purpose of this paper is to process the field of face detection. In today's world it forms an indispensable part in - social networks or security systems. It mainly utilizes thematic recognition based on renowned features of HOG, Haar and LBP. The aim of the following section is to investigate and describe the rules of how these features function. In addition,the paper will carry out an implementation of face detection with the use of cascade classifier that is tailored to the features mentioned above. For the implementation the library OpenCV 2.4.8 will be used. In terms of experimental testing of individual implementations, a separate testing set of images is prepared. It will be used to perform a test of speed, accuracy and functionality of these detectors. The paper will conclude with performing a comparison analysis of available data.