Abstract

Visual tracking is one of the central problems in computer vision. A crucial problem of tracking is how to represent the ob ject. Tra- ditional appearance-based trackers are using increasingly more complex features in order to be robust. However, complex representations typi- cally will not only require more computation for feature extraction, but also make the state inference complicated. In this paper, we show that with a careful feature selection scheme, extremely simple yet discrimi- native features can be used for robust ob ject tracking. The central com- ponent of the proposed method is a succinct and discriminative repre- sentation of image template using discriminative non-orthogonal binary subspace spanned by Haar-like features. These Haar-like bases are se- lected from the over-complete dictionary using a variation of the OOMP (optimized orthogonal matching pursuit). Such a representation inherits the merits of original NBS in that it can be used to efficiently describe the ob ject. It also incorporates the discriminative information to distin- guish the foreground and background. We apply the discriminative NBS to ob ject tracking through SSD-based template matching. An update scheme of the discriminative NBS is devised in order to accommodate ob ject appearance changes. We validate the effectiveness of our method through extensive experiments on challenging videos and demonstrate its capability to track ob jects in clutter and moving background.