Silhouettes have been used extensively for human action recognition. Using a low resolution, this bidimensional data is sometimes employed as it is, for instance, relying on a volume of silhouettes as spatio-temporal feature. These binary masks can then be reduced in dimensionality as in [19], where principal component analysis (PCA) is employed. TT propose a log-polar histogram which is computed choosing the different radii of the bins based on logarithmic scale. A bag of rectangles has been proposed in [9], where a histogram of oriented rectangular patches is extracted over the whole silhouette. Similarly, in [23], a 3D histogram of oriented gradients (3DHOG) is extracted on densely distributed regions. A very popular feature is the one from Tran and Sorokin [20]. It combines silhouette shape and optical flow in the same feature vector. By means of radial histograms the silhouette shape and the X and Y axis optical flow are encoded and combined with the context of 15 surrounding frames. In [7], the authors present a low-dimensional radial feature that is based on the contour points of the silhouette, and shows suitability for real-time processing. Since the silhouette can be easily obtained as a binary mask using the depth data provided by the Microsoft Kinect device, all these silhouette-based features can also be applied on RGB-D images