Image segmentation and object tracking method and corresponding system

Abstract

This relates to a method for image segmentation and video object tracking. While conventional methods rely on the motion homogeneity of the objects to be tracked, the present method is based on a concept of partition projection in which both spatial and motion information are now exploited, and implemented by means of a modified double partition approach, including a re-segmentation of a first partition followed by a projection step of the obtained fine partition into the current frame. Objects presenting non-homogeneous motion, static objects, and parts of an object with homogeneous motion can thus be tracked.

Description

technical field [0001] The present invention relates to the object tracking method that is applicable to image sequence, described method comprises the following steps in sequence: [0002] (A) Segmentation step, which determines the region of the first coarse partition P(t-1) of the image I(t-1), said region corresponds to the object being tracked, and in said coarse partition P( On the basis of t-1), the subdivision FP(t-1) is determined according to the principle of spatial uniformity; [0003] (B) a projection step, which determines the projection PFP(t) of the subdivision FP(t-1) in the current image I(t); [0004] (C) Relabeling step of said projected subdivision PFP(t), which determines the final partition P(t). [0005] The invention also relates to a system corresponding to implementing said method. Background technique [0006] Cosine transform coding is one of the most famous compression techniques in image data transmission, but its compression ratio cannot ex...

AI Technical Summary

Problems solved by technology

[0009] Traditional object tracking technology uses motion as the main information. When several local parts that make up the object show different motions (such as a walking person whose arms and body are doing different motions), it may lead to tracking failure.

In addition, motion-based tracking techniques cannot track parts of an object when the entire object is doing a given motion (for example, the technology cannot separate from the human hair and only track the human face)

Finally, if the entire scene is static (objects are not moving) or if the camera has an overall motion (such as panning), motion-based tracking techniques will not be able to track or will have difficulty tracking selected objects

Static scenes (or scenes that become static) do not provide any motion information, so it is actually difficult to detect objects based on their motion

Similarly, the overall motion of the camera makes all objects in the scene have an obvious motion, so it is not easy to distinguish between static and moving areas to detect objects

Images