| In the television broadcasting system, television signals will still use interlaced formatfor a long time considerations for compatibility across different television and the cost oftransmission bandwidth. For high-definition digital television(HDTV), it only supportprogressive scanning technology, and the direct display of interlaced video on these HDTVwill cause serious line flicker, line crawl, serration and parallel phenomenon and so no.Thus, de-interlacing is necessary steps before the TV image was input to the displaysystem.Main work of this dissertation is to study the de-interlacing algorithm based on motionadaptive method, and try to use the correlation of the fields to detect the static and motionregions in the field, and classify the edges in the field to improve the interpolation accuracyof the video so as to reach the aim of improving the de-interlacing efforts.This research provides some new insights into de-interlacing algorithm:(1) Since the incomplete information of the field was used during the motion detection,and then it is easy to cause the error detection, based on the traditional four-field motiondetection method, a correlation of fields based motion detection algorithm is designed inthis dissertation. Firstly, this algorithm halves the original image to form a complete frame,and then according to the correlation coefficients for two adjacent fields with the sameparity, we split the field into moving regions and stationary regions. Empirical results showthis method performs well and improves the accuracy of detection.(2) Traditional intra-field interpolation algorithm mainly proposed for wide edges, andit is difficult to calculate the direction of the narrow edges. In order to improve theprocessing effort for the edge regions, by analyzing the edge based line average method,this dissertation subdivide the edges and presents a new Edge classification fieldinterpolation algorithm. The experiments verify the feasibility and effectiveness of thismethod. |
PDF Full Text Request