Research And Realization Of Moving Target Tracking Algorithm Based On Video

As long as the second Internet Technology revolution waves swept over the whole World, the study enthusiasm of scholars and experts in information science are rising.Research on computer vision, especially the kernel technology-visual tracking has achieved significant process. Visual tracking has a wide range of applications prospect on Intelligent surveillance,Human-computer interaction, Medical image processing. At the same time, it plays an irreplaceable role in the civil and commercial aspects. In recent decades, experts both inside and outside of China in related field continue to innovation, many state-of-art algorithms have been proposed.However, due to the constraints of many factors, such as illumination variation, appearance change, partially or fully occlusion, and the complex background and so on,vision tracking in practical application still exist many difficulties. Therefore, design a robust real-time vision tracking algorithm is a very challenging task.Based on the achievements of previous researches, this paper has carried on the video target tracking algorithm research and experimentation, the following aspects were discussed on the study:1. Online weighted multiple instance object tracking via active feature selectionAdaptive tracking-by-detection methods have been widely used in computer vision. These approaches treat the tracking problem as a classification task and use online learning techniques to update the object model. To solve the common problem of adaptive object tracking —“drifting”, an online weighted multiple instance object tracking via active feature selection is proposed, in the prime framework of multiple instance learning,we draw a new kind of weighted bag probability model —Weighted sum model, and then use an active feature selection approach to select more informative features to decrease the uncertainty of classification model,otherwise, We optimize the Fisher information criterion to select features in an online boosting method.2. The Spatio-Temporal Context learning vision tracking algorithm via MFFTAlong with the computer vision technology application scenario is rich, people has higher requirement for tracking algorithm performance. So, a simple yet fast and robust algorithm which exploits the MFFT for visual tracking is proposed. This approach formulates the Spatio-temporal relationships between the object of interest and its local context based on a Bayesian framework, which models the statistical correlation between the low-level features from the target and its surrounding regions. The tracking problem is posed by computing a confidence map, and obtaining the best target location by maximizing an object location likelihood function. Meanwhile, a modified Fast Fourier Transform is adopted for fast learning and detection in this work, make the tracking algorithm is more efficient.