There are many motivations for using features rather than pixels
directly. For mobile robots, a critical motivation is that
feature-based systems operate much faster than pixel-based
systems [25]. The features used here have the same
structure as the Haar basis functions, i.e., step functions
introduced by Alfred Haar to define wavelets
[8]. They are also used in
[12,15,25].
Fig. (left) shows the eleven basis features, i.e.,
edge, line, diagonal and center surround features. The base
resolution of the object detector is for instance
pixels, thus, the set of possible features in this area is very
large (642592 features, see [12] for calculation
details). In contrast to the Haar basis functions, the set of
rectangle features is not minimal. A single feature is
effectively computed on input images using integral images
[25], also known as summed area tables
[12]. An integral image is an intermediate
representation for the image and contains the sum of gray scale
pixel values of image with height and width , i.e.,
|
To detect a feature, a threshold is required. This threshold is automatically determined during a fitting process, such that a minimal number of examples are misclassified. Furthermore, the return values of the feature are determined, such that the error on the examples is minimized. The examples are given in a set of images that are classified as positive or negative samples. The set is also used in the learning phase that is briefly described next.