Recearch On Algorthm Of Digital Video Intelligent Edittng

During producing video, in order to improve the visual effect, usually, it is more likely to remove the unexpected text, or spot in the image, for example the pots on people's faces. Traditionally, we need to re-film that video clips, or use Photoshop to remove the unexpected object from the video frame by frame manually. That will expand the produce period quite long, and increase the expense intensively. In order to overcome these short nesses, this research designs a Digital Video Intelligent Edit System to edit the flaws on video images, besides the system framework, this research emphasizes on researching the object-tracking algorithms and digital image restore algorithms.Firstly, in order to scale down the project size to an acceptable range, this paper focuses on processing micro subject, and assumes that the transfer and movement of the subject is continuous. That means the area of the target is limited, and the target only contains little texture information, it makes difficult to extract feature point from the target. Therefore, object-tracking algorithm based on regional is the most appropriate one. The background of the micro target usually contains rich texture information and feature, and it can be used to assist tracking the object. Take all factors into consideration, this paper proposes an object-tracking algorithm that combines KLT and block matching algorithm together; the research firstly applies KLT to extract the affine coefficient matrix, and recalculates the micro target block model in current frame by applying the affine coefficient matrix on the micro target in previous frame to get the new micro target model, and finally applies blocking matching to track the exact position of the micro target.Secondly, FMM algorithm not only has good performance in time complexity, but also can restore the image with good result, this paper applies Telea FMM impainting algorithm to implement the restore module. Since the target that to be restoration is in video not only for one frame, this paper proposes a method to improve Telea FMM algorithm, the improved algorithm adaptively changes the neighborhood size based on macro target size and affine transfer factor which is calculated by KLT algorithm mentioned previous, and that improves the restoration quality to better and stable level.Finally, this paper designs a Digital Video Intelligent Edit System framework based on object-tracking and image restoration, and at the same time adds User interaction to provide the user enter related parameters. The system applies object-tracking algorithm to track the object in the continuous frames, and then applies image restoration algorithm to remove these unexpected objects from the video. It refers the functions of Photoshop and innovatively add object-tracking module to convert it from image edit system framework to video intelligent edit system framework.