Vison-based Blank Paper Tracking Algorithm Research

In recent years,vision tracking has been undergone rapid development.And it has been widely used in video surveillance,human-computer interaction and other fields.At present,many tracking algorithms are present for moving object tracking,such as the object tracking with low resolution,scale changes,fast moving,the problems of illumination change in the process of object tracking and the improvement of tracking rate are solved better.But most of these scenes are rich in texture detail.Different from the existing research work,this paper takes the blank paper as the object for tracking,and realizes the accurate positioning of the paper by object tracking.For blank paper,unlike other natural scenes,since the lack of texture characteristics of the blank paper in the office scene,and there is often a large area of occlusion by hand or other objects.In addition,a large range of motion of the paper object is optional and there is a large rotation.Therefore,it is a strong challenge to track the blank paper.This paper mainly includes the following two aspects:(1)For blank paper object,several common target tracking algorithms and their applications are analyzed and compared.It is found that the existing algorithms are prone to drift and loss.(2)Traditional vision tracking algorithms cannot track blank paper very well since blank paper lacks significant texture details leading to insufficient feature points.And the situation becomes worse when the tracking object moves with large rotation.To solve these problems,an improved compressive tracking algorithm with structure constraint is proposed.Firstly it uses line detection to find the line segments of the tracking samples,and then measure the angles among the found line segments to remove those obvious wrong samples with similar feature values.On one hand,the structure constraint reduces the sample search space and enables the compressive tracking algorithm to track objects with large rotation.On the other hand,the four exact corners of the blank paper are tracked at the same time,which are used to locate the position of the whole paper.The experiments have fully confirmed the validity of proposed compressive tracking algorithm with structure constraint for blank paper visual tracking.Moreover,illustrating it can track well for such blank paper without texture when the object undergoes large rotation.