Object-based detection of vehicles in airborne data

Abstract

Robust detection of vehicles in airborne data is a challenging task since a high variation in the object signatures - depending on data resolution - and often a small contrast between objects and background lead to high false classification rates and missed detections. Despite these facts, many applications require reliable results which can be obtained in a short time. In this paper, an object-based approach for vehicle detection in airborne laser scans (ALS) and photogrammetrically reconstructed 2.5D data is described. The focus of this paper lies on a robust object segmentation algorithm as well as the identification of features for a reliable separation between vehicles and background (all none-vehicle objects) on different scenes. The described method is based on three consecutive steps, namely, object segmentation, feature extraction and supervised classification. In the first step, the 2.5D data is segmented and possible targets are identified. The segmentation progress is based on the morphological top-hat filtering, which leaves areas that are smaller than a given filter size and higher (brighter) than their surroundings. The approach is chosen due to the low computational effort of this filter, which allows a fast computation even for large areas. The next step is feature extraction. Based on the initial segmentation, features for every identified object are extracted. In addition to frequently used features like height above ground, object area, or point distribution, more complex features like object planarity, entropy in the intensity image, and lineness measures are used. The last step contains classification of each object. For this purpose, a random forest classifier (RF) using the normalized features extracted in the previous step is chosen. RFs are suitable for high dimensional and nonlinear problems. In contrast to other approaches (e.g. maximum likelihood classifier), RFs achieves good results even with relatively small training samples

Improving semantic orthophotos by a fast method based on harmonic inpainting

“…As mentioned before, we have as the main input an orthophoto in which vehicles were detected by one of the standard procedures, such as a method based on boosting [3] or a regional-based approach [14]. In this latter method, also elevation data was used to compute features for classification.…”

Segmentation methods for detection of stationary vehicles in combined elevation and optical data

“…D1: [Top-hat operator] The initial segmentation of the procedure proposed by [15] implied application of the tophat filter on the elevation map. The top-hat transformation is actually a thresholded difference between the image and its morphological opening with a given filter size.…”

“…The typical input for any semantic urban terrain reconstruction procedure is a DSM and an orthophoto; both may stem from airborne laser scanning or from the 3D reconstruction given aerial images [14]. At each case, as [15] pointed out, elevation data is an extremely useful source of information, which can and should be extensively exploited. Our second difference with respect to [10] is that we are interested in improving the quality of segmentation of cars without constraints on orientation, vicinity to a-priori estimated street course, etc.…”

Reconnaissance of coastal areas using simulated EnMAP data in an ERDAS IMAGINE environment

“…Remote sensing technologies are already widely used in military context [2,8,18]. Flying platforms and their corresponding sensor-(systems) experienced a fast development during the last decade [10,13,16], mainly driven by the establishment of drone technology, active 3D point measurements and the increasing number of operating sensor channels. Passive sensors with the highest number of imaging bands are the so-called hyperspectral (HS) sensors or spectrometers.…”