Instrumentation and Algorithms for the Identification of Ill-formed Objects

The general area of research described here relates in part to computational pattern recognition and image processing. More specifically however, it relates to image recognition of amorphous objects, which due to their ill-defined shape, are difficult to identify. In particular, the focus for the research relates to vehicle mounted weather condition sensors that are operated during journeys on public roads and how these sensors can be used to gather sufficiently comprehensive data for algorithmic analysis such that timely and accurate results can be obtained. In large part this work has been conducted together with researchers in the Connected Car Project, based at BMW in Munich, Germany. The principal research question is how accurately data can be obtained from vehicle-mounted cameras and LIDAR (Light Detection and Recognition) to identify ill-formed objects such as fog, low cloud and particulate matter such as smoke. The operation of the sensor equipment, building of a data matrix for analysis, algorithm design and the results from experiments with these instruments is the context for this paper. Keywords-Image Processing, Pattern Recognition, Algortihm Design, Air Pollution Detection, Fog Detection, Weather Detection, Remote Sensing, LIDAR, Colaborative Driver Assistant Functions, Spatial Resolution, Air Pollution Forecasting Services.