The Goal Of A Complex Ground Background Real-time Tracking Algorithm

Object Real-time Tracking Arithmetic is one of the most important applications in graphics transaction area. This arithmetic provides a sequence tracking method to track original object as a standard image. During the tracking process it will control the series states of the standard image such as changing image, judging lost image and etc. There is a great contrast between tracking the object from a sample background and tracking it from the real complex ground background. It is so hard to pick-up object's various characteristics for example its shape. So there are many differences between them. There is various tracking arithmetic. Most of them are based on image matching method. Its academic base is comparability measurement.In this paper, after researching various arithmetic of real-time tracking the author probe into these correlative tracking arithmetic and provide a new real time tracking arithmetic based on template self- correlation. It is a useful arithmetic that proved effective and feasible by series examination.In the first part of the paper, the author introduces many multi- templates tracking arithmetic in the graphics tracking area. Then based on these arithmetic, the author deliver a new Five- template correlation method to calculate the correlative result reliability by the position of the object and the background. At the same time an improved algorithm which can calculate the correlative position is provided in this thesis. This algorithm use one "lost-matching matrix" and one "weight class distributing matrix" to calculate the template-related correlative result reliability and object's position, calculate the steady tracking probability about the absolute correlative gap. Compare with the single-template tracking arithmetic and the improved multi- template tracking arithmetic, the result of the improved multi-template tracking arithmetic is much better than the single-template tracking arithmetic.Then the author analysis the pertinence of the image , provide a new "distilling template algorithm" which was based on image pertinence and more judgment guide line in this improved multi- template tracking arithmetic, such as replacing template guide line, losing object judgment guide line and etc.At last the author introduces a laboratorial real-time tracking system. This part includes the introduction about each component of the system and the work flow of...