The HOG method is applied in the detection of human bodies, specially when they are in a vertical position and in many backgrounds. HOG method was evaluated before in different applications such as pedestrian detection, video surveillance, search and rescue. However, when human bodies are in other positions, most of the time, body recognition algorithms present fails. The main idea presented in this research, is the evaluation of different morphological operators applied to improve the HOG method. These experiments show that the results of combining HOG method with morphological operators are better than just using the HOG method. In this research the HOG method combined with morphological operator close (86, 62%) and Erode (84, 35%) had better results than HOG without this pre-processing (77, 32%).