Particle Filter with Integrated Multiple Features for Object Detection and Tracking
Muhammad Attamimi, Takayuki Nagai, Djoko Purwanto
Considering objects in the environments (or scenes), object detection is the first task needed to be accomplished to recognize those objects. There are two problems needed to be considered in object detection. First, a single feature based object detection is difficult regarding types of the objects and scenes. For example, object detection that is based on color information will fail in the dark place. The second problem is the object’s pose in the scene that is arbitrary in general. This paper aims to tackle such problems for enabling the object detection and tracking of various types of objects in the various scenes. This study proposes a method for object detection and tracking by using a particle filter and multiple features consisting of color, texture, and depth information that are integrated by adaptive weights. To validate the proposed method, the experiments have been conducted. The results revealed that the proposed method outperformed the previous method, which is based only on color information.
object detection; object tracking; multiple features; features integration; particle filter;