Tarek El-Gaaly - Final Thesis Document
نویسندگان
چکیده
Particulate Matter (PM) is a form of air pollution that visually degrades urban scenery and is hazardous to human health and the environment. Current monitoring devices are limited in measuring average PM over large areas. Quantifying the visual effects of haze in digital images of urban scenery and correlating these effects to PM levels is a vital step in more practically monitoring our environment. Current image haze extraction algorithms remove haze from the scene for the sole purpose of enhancing vision. We present two algorithms which bridge the gap between image haze extraction and environmental monitoring. We provide a means of measuring atmospheric scattering from images of urban scenery by incorporating temporal knowledge. In doing so, we also present a method of recovering an accurate depthmap of the scene and recovering the scene without the visual effects of haze. We compare our algorithm to three known haze removal methods. The algorithms are composed of an optimization over a model of haze formation in images and an optimization using a constraint of constant depth over a sequence of images taken over time. These algorithms not only measure atmospheric scattering, but also recover a more accurate depthmap and dehazed image. The measurements of atmospheric scattering this research produces, can be directly correlated to PM levels and therefore pave the way to monitoring the health of the environment by visual means. Accurate atmospheric sensing from digital images is a challenging and under-researched problem. This work provides an important step towards a more practical and accurate visual means of measuring PM from digital images.
منابع مشابه
A Comparative Analysis and Study of Multiview CNN Models for Joint Object Categorization and Pose Estimation Supplementary Materials
متن کامل
Factorization of view-object manifolds for joint object recognition and pose estimation
Due to large variations in shape, appearance, and viewing conditions, object recognition is a key precursory challenge in the fields of object manipulation and robotic/AI visual reasoning in general. Recognizing object categories, particular instances of objects and viewpoints/poses of objects are three critical subproblems robots must solve in order to accurately grasp/manipulate objects and r...
متن کاملMeasuring Atmospheric Scattering from Digital Image Sequences
Current environmental monitoring devices are limited in their capability of measuring atmospheric particulate matter (PM) over large areas. Quantifying the visual degrading effects of atmospheric scattering in digital images of urban scenery and correlating these effects to PM levels is a vital step in more practically monitoring our environment. Currently, image haze removal (or dehazing) tech...
متن کاملMulti-Modal RGBD Sensors for Object Grasping and Manipulation
RGBD sensors, such as the Microsoft Xbox Kinect [1] are types of multi-modal perceptual sensors that have appeared in recent years. RGBD sensors have become standard perceptual tools for robots as they provide a unique multi-modal approach to perception. A vital pre-cursing challenge in object grasping and manipulation is object pose recognition. A robot must identify the pose (i.e. orientation...
متن کاملVisual Obstacle Avoidance for Autonomous Watercraft using Smartphones
This paper presents a visual obstacle avoidance system for low-cost autonomous watercraft in riverine environments, such as lakes and rivers. Each watercraft is equipped with a smartphone which offers a single source of perceptual sensing, via a monocular camera. To achieve autonomous navigation in riverine environments, watercraft must overcome the challenges of limited sensing, low computatio...
متن کامل