Thermal 3D mapping for object detection in dynamic scenes

Abstract

ABSTRACT:The automatic analysis of 3D point clouds has become a crucial task in photogrammetry, remote sensing and computer vision. Whereas modern range cameras simultaneously provide both range and intensity images with high frame rates, other devices can be used to obtain further information which could be quite valuable for tasks such as object detection or scene interpretation. In particular thermal information offers many advantages, since people can easily be detected as heat sources in typical indoor or outdoor environments and, furthermore, a variety of concealed objects such as heating pipes as well as structural properties such as defects in isolation may be observed. In this paper, we focus on thermal 3D mapping which allows to observe the evolution of a dynamic 3D scene over time. We present a fully automatic methodology consisting of four successive steps: (i) a radiometric correction, (ii) a geometric calibration, (iii) a robust approach for detecting reliable feature correspondences and (iv) a co-registration of 3D point cloud data and thermal information via a RANSAC-based EPnP scheme. For an indoor scene, we demonstrate that our methodology outperforms other recent approaches in terms of both accuracy and applicability. We additionally show that efficient straightforward techniques allow a categorization according to background, people, passive scene manipulation and active scene manipulation.

Superimposing Thermal-Infrared Data on 3D Structure Reconstructed by RGB Visual Odometry

“…Matsumoto et al [4] uses a hair dryer to heat the calibration board while Ham et al and Weinmann et al [10], [13] prefer to use a set of lamps. Some other methods proposed by Prakash et al and Saponaro et al [14], [15] use a flood lamp to heat the calibration board.…”

Concept for Classifying Facade Elements Based on Material, Geometry and Thermal Radiation Using Multimodal Uav Remote Sensing

“…However, since the checkerboard pattern is not visible in the spectrum of thermal infrared, it is not possible to utilize regular checkerboard patterns for thermal sensors. Geometric calibration of thermal sensors is therefore often done by utilizing a pattern consisting of lamps (Lagüela et al, 2011a, Weinmann et al, 2014, Ellmauthaler et al, 2013. By creating a regular grid of lamps, the possibility of using standard calibration procedures for sensors is enabled.…”

“…A common approach to generate 3D models of facades is with the use of a laser scanner, which can be merged with thermal data to generate thermal 3D models (Weinmann et al, 2014, Borrmann et al, 2013. However, since a laser scanner will not be used for this concept, another method has to be approached.…”

Registration of RGB and Thermal Point Clouds Generated by Structure From Motion

“…A similar method is proposed in [7], where the sensors are instead placed on a wearable backpack comprising of five 2D laser range scanners, two optical cameras and two infrared cameras. In [26], the authors present a registration method using a range camera that is able to simultaneously provide both range and intensity images. Given 2D correspondences between the intensity and thermal images, the range camera can assign a 3D point to each 2D matches.…”