Research And Implementation Of An Automatic Keying Technology

Video after effect editing is using computer graphics to do some digital conversion on video frames, so as to achieve some kind of visual and even incredible screen effect. Traditional video editing, mainly applied in professional fields, is hard to be popularized as a result of massive video editing data, processing complexity and high resource consumption. With the development of digital technology and the improvement of hardware, image and video after effect is no longer mysterious and more and more relevant technologies come out and entered ordinary families from laboratories. The perfect combination of art and technology turns the hard into the possible, adventurous and exciting and create dreamlands one after another, which is absolutely a dream factory.Nowadays, the requirements of video after effect editing in non-professional application fields, for example home entertainment, are ever-growing, while the traditional technologies are difficult to popularize and apply.This paper studies on video after effect editing technology, which presented a whole set solution of video capture, video edit and play, mean while implemented easy operation and automatically real time video after effect editing. Video after effect editing processes large amount of data, which makes the efficiency to be an urgent issue. This paper is based on pipeline principle and combined with powerful computer vision lib OpenCV and raised real time after effect loading and rendering solution. All kinds of video effects are presented by filters, running by pipeline, and implemented real time automatic video after effect editing. It achieved common real time rendering, such as blurring, aged film and subtitle loading.Image matting is the basis and key technique of digital effect, also take as one of the most complicated technique. Our paper combined with digital image processing and computer vision related knowledge and presented automatically, low background limitation and operation easily digital matting algorithm. Traditional natural image matting algorithm needs quite a lot user interaction, it must provide trimap manually, which makes frequent interaction and poor real-time. This paper studies in depth on traditional digital image matting algorithm, combined with image saliency theory and segmentation technique and raised an automatic natural image matting algorithm.In the first place, it got saliency map by using image saliency theory, and then got rough partition of foreground and background after binary segmentation. Finally after eroding and dilating, and got a set of complete trimap. In consideration of the errors in automatically generated trimap and much roughness in unknown space, this paper inspected highly visual consistency of color similarity measurement of CIE1976L*a*b*color space, expanded known space, improved the traditional RGB space based Shared Matting algorithm, and got better image matting effect which improved the efficiency of the algorithm..