Estimation of human body height from uncalibrated image

In this paper the method for extraction of selected human body features, particularly dimensional units, is presented. The key parameters we focus on are the total human body height, arm length, leg length, shoulders distance and head height. Prototype software solution utilizes Histogram of Oriented Gradients (HOG) and Haar classifiers algorithms. The analysis of existing object detection approaches in the image is carried out. Following this analysis, the suitable approach is selected, while bearing in mind the software prototype requirements. Further, the methods for extraction of human body parameters are presented and specific details of implementation are described. Prototype of the application enables specification of selected human body features having the output values close to the real dimensions of human body.