Movement Classification is the process of classifying and recognizing observed motion. In the human visual system this process plays an important role. For instance, humans can recognize others by his/her gait, hand gestures and other complex sequences of motion. A lot of other animals are also believed to make use of this process, for instance for recognizing prey or predator. Recently, a lot of effort has been put into applying movement classification on visual data. Traditionally, there are two methods used to analyze motion in this data. The first, Structure from Motion (SFM) tries to build structure from a series of frame, resulting in a time set of coordinates of points found in these frames. This derived structure is then used for classification and recognition. Unfortunately a lot of other useful and sometimes vital cues for recognizing motion are disregarded in this way, like optical flow and textures. (Dijk, 01)
Several terms describe how and where an animal moves. Aquatic animals swim; volant animals fly. Cursorial animals (cursors) run rapidly and for long distances. Scansorial animals are climbers; in the extreme, they are arboreal, spending most of their lives in the trees. Hoppers are termed saltatorial. If they use their hindlimbs only and in a fast succession of hops, they are said to be ricochetal. Fossorial forms are diggers, usually living in burrows. Here, we focus on the adaptations of cursors.
A full cycle of motion of a running or walking mammal is called a stride. An animal's speed is the product of its stride length times rate. There are two ways of increasing the speed of running, increasing stride length and increasing stride rate. Some animals are clearly specialized to increase speed through increasing stride length; the giraffe is an extreme example.........
