Fast object segmentation from a moving camera

Abstract

Segmentation of the scene is a fundamental component in computer vision to find regions of interest. Most systems that aspire to run in real-time use a fast segmentation stage that considers the whole image, and then a more costly stage for classification. In this paper we present a novel approach to segment moving objects from images taken with a moving camera. The segmentation algorithm is based on a special representation of optical flow, on which u-disparity is applied. The u-disparity is used to indirectly find and mask out the background flow in the image, by approximating it with a quadratic function. Robustness in the optical Row calculation is achieved by contrast content filtering.The algorithm successfully segments moving pedestrians from a moving vehicle with few false positive segments. Most false positive segments are due to poles and organic structures, such as trees. Such false positives are, however, easily rejected in a classification stage. The presented segmentation algorithm is intended to be used as a component in a detectiodclassification framework.

Sensors for Obstacle Detection - A Survey

“…The extraction of depth information (also called stereo matching) is a process of finding correspondences between two correlated images [1], [2]. The main limit of the motion based segmentation technique is the need of processing a number of successive frames which increases the processing time [3]. The information extracted after segmentation can be a model or some features of the object.…”

Generic fusion of visual cues applied to real-world object segmentation

“…A common mode of motion in the image is found by applying u-disparity, and any region in the image that exhibits a different motion will be segmented. The details of this algorithm have previously been published in [14].…”

“…It can in general not find non-moving objects (as discussed in [14]). However, the algorithm does find nonmoving objects that significantly stand out from the background, due to the camera motion.…”

Real-Time Foreground Segmentation for the Moving Camera Based on H.264 Video Coding Information

“…Motion-based segmentation techniques extract regions with homogeneous motion vectors (MVs) and achieve fast segmentation [4], [5]. These methods treat the video compression as a kind of pre-processing, as the video coding is inevitable in most of video applications.…”