| Popis: |
In computer vision, interest point tracking across an image sequence is a fundamental technique for determining motion characteristics. Once motion is extracted it can be applied to a wide array of other practical applications, including but not limited to automated surveillance, three-dimensional recovery, and object recognition. Traditionally, feature tracking has been performed with a variety of appearance-based comparison methods. The most common methods analyze intensity values of local pixels and subsequently attempt to match them to the most similar region in the following frame. This standard, though sometimes effective, lacks versatility. For example, these methods are easily confused by shadows, patterns, feature occlusion, and a variety of other appearance-based anomalies.To counteract the issues presented by a one-sided approach, a new method has been developed to take advantage of both appearance and geometric constraints in a complementary fashion. To do this, a select number of points are first tracked through a set number of initialization frames such that their shape can be defined. Beginning at the following frame, this spatial information is substituted as an additional constraint into the appearance-based optical flow equation. Through aniterative least-squares solution the camera parameters for the new frame are computed and used to project the derived shape data to the new feature point image coordinates. The process is repeated for each new frame until a trajectory is created for the entire video sequence.With this method, weight can be allocated as desired between both appearance and geometric constraints. If an issue arises with one constraint (e.g., occlusion or rapid camera movement), the other constraint will continue to track the feature successfully. Preliminary results have shown that this method provides consistent robustness to tracking challenges, such as occlusion, shadows, and repeating patterns, while also outperforming appearance-based methods in tracking quality. With this improvement, many existing deficiencies in the practical applications of feature tracking can eventually be overcome. |
