Using Human Objects for Illumination Estimation and Shadow Generation in Outdoor Environments

Abstract

In computer graphics and augmented reality applications, the illumination information in an outdoor environment enables us to generate a realistic shadow for a virtual object. This paper presents a method by which to estimate the illumination information using a human object in a scene. A Gaussian mixture model, in which the mixtures of Gaussian distributions are symmetrical, is employed to learn the background. The human object is then segmented from the input images and the disparity map obtained by a stereo camera. The ground plane in the scene, which is important for estimating the location of the human object on the ground, is then detected using the v-disparity map. The altitude and the azimuth value of the sun are computed from the geometric relationship of three scene elements: the ground, human object, and human-shadow region. The experimental results showed that the proposed method can estimate the sun information accurately and generate a shadow in the scene for a virtual object.

Anthropometric Ratios for Lower-Body Detection Based on Deep Learning and Traditional Methods

“…A human bounding box can be detected by a refinement algorithm by matching the contour of the shadow of the human body. In addition, the Gaussian mixture model (GMM)-based background learning technique was applied to separate the human object from the background [33]. Iazzi et al [34] also applied the background subtraction with a support vector machine (SVM) classifier to detect a fall in elderly people.…”