Provided are an apparatus and a method for reliably detecting a human being from an input video image by using tilt-based feature vectors and neural networks. The provided apparatus comprises an image preprocessing unit for modeling a background image from an input image; a motion object region setting unit for differentiating the input image and the background image to set a motion object region in which a motion exists; a human being region detecting unit for extracting feature vectors of the whole body and the upper body based on a tilt from the motion object region, and inputting the extracted feature vectors of the whole body and the upper body based on a tilt to a neural network classifier to detect a human being region; and a determining unit for determining whether or not an object in the detected human being region is a human being. By using tilt-based feature vectors and neural networks having excellent discernment with respect to the whole body and the upper body of various types of images, a human being can be detected reliably.
申请公布号
KR20140081254(A)
申请公布日期
2014.07.01
申请号
KR20120150808
申请日期
2012.12.21
申请人
ELECTRONICS AND TELECOMMUNICATIONS RESEARCH INSTITUTE
发明人
LIM, KIL TAEK;CHUNG, YUN SU;HAN, BYUNG GIL;CHOI, EUN CHANG;LEE, SOO IN