Automatic Person Detection and Tracking using Fuzzy Controlled Active Cameras

Abstract

This paper presents an automatic system for the monitoring of indoor environments using pan-tilt-zoomable cameras. A combination of Haar-feature classifier-based detection and color histogram filtering is used to achieve reliable initialization of person tracks even in the presence of camera movement. A combination of adaptive color and KLT feature trackers for face and upper body allows for robust tracking and track recovery in the presence of occlusion or interference. The continuous recomputation of camera parameters, coupled with a fuzzy controlling scheme allow for smooth tracking of moving targets as well as acquisition of stable facial closeups, similar to the natural behavior of a human cameraman. The system is tested on a series of natural indoor monitoring scenarios and shows a high degree of naturalness, flexibility and robustness.

Intelligent active vision system for autonomous robots

“…Nowadays, several simple fuzzy controllers are proposed for camera position control, among which we mention the condensed 4-rules controller, generally presented in [17] and concretely used in [4]. More complex rule-bases are proposed in [2] (with 9 fuzzy rules) and in [3] (with 15 and 25 fuzzy rules). The condensed PD-like fuzzy controller is simple enough for an easy implementation with satisfactory performances.…”

Person re-identification with a PTZ camera: An introductory study

“…PTZ cameras are nowadays widespread in many different environments (stadiums, banks, crossroads) and many tracking algorithms have been developed to automatically zoom on particular areas of interest [11,12,13]. Assuming this technology available, we present a PTZ protocol composed by a set of actions (zoom on the head, torso, legs, do not zoom), which produces a two-image description of each probe subject.…”

PTZ Camera Modeling and Panoramic View Generation via Focal Plane Mapping

“…Similarly [4] employs a search mechanism to minimize an SSD matching score function, which again is computationally expensive. In [3], a correction matrix is used to continuously update the camera parameters with the latest pan and tilt information as the camera moves. However, this information is not precise enough for pixel-level mapping.…”

“…Although various complex camera models have been proposed for these tasks [1][2][3][4][5], we expand on the preliminary work of [6] and propose a real-time, reliable and completely automatic PTZ mapping model. We theoretically derive the minimum formulation necessary to map any arbitrary point on the focal plane of a PTZ camera to its corresponding pan-tilt orientation, based on the current pantilt value of the camera.…”