| With the development of science&technology and the improvement of livingstandards, people hope for the higher and higher quality of TV, from ordinary analogTV to digital TV which is pervasive at present, from black-white TV with simplefunction to color TV with multimedia technology, from CRT TV to LCD TV.Although the technological progress of TV improves dramatically, but due to variousfactors, the analog signal will not disappear immediately, when traditional interlacedsignal displaying on a digital flat-panel TV, crawling and flicker will be moreseriously, so the technology of deinterlacing particularly important. This thesis mainlyfocuses on research of deinterlacing.In this thesis, we make the introduction and analysis about the commondeinterlacing algorithm. Today, deinterlacing algorithm includes linear deinterlacingalgorithm, nonlinear algorithm, motion estimation and motion compensation. Lineardeinterlacing algorithm includes spatial filter, temporal filter and spatial&temporalmixed filter. Spatial filter is based on the information of the current field to interpolate,such as line copy method and line averaging method. Temporal filter is based on theinformation of the adjacent fields to interpolate. Mixed spatial and temporal filter isbased on the information of time and space to interpolate. Nonlinear deinterlacingalgorithm includes motion adaptive algorithm, boundary adaptive algorithm andmedian filter. Motion adaptive algorithm is based on the motion information of imageto select the current field or the adjacent field to interpolate. Boundary adaptivealgorithm is based on the boundary information of the object in the image tointerpolate. In median filter algorithm, interpolated pixel point is the intermediatevalue of the neighbor pixels in the current and adjacent fields. The motion estimationalgorithm is based on the information of adjacent fields to estimate motion vector of the image. The motion compensation algorithm interpolate according to the motionvector of the image.Due to the traditional deinterlacing algorithm searches bigger angle and smallerrange, this thesis presents a new deinterlacing algorithm which is based on edgedetection. Through the luminance information of the image to judge the edgeinformation of the object in the image, then detects to find the pair of pixel which isthe most relevant, taking the average of the two pixels as the interpolater pixel. Thenthrough mean filtering to gain the image which is line by line.Based on the above algorithm, this thesis proposes the hardware implementationof the deinterlacing algorithm. In this thesis, how to make the hardwareimplementation running in a faster rate and occupying a fewer space are importantand difficult. To improve processing speed and reduce the occupied hardwareresource through reusing linebuffer, computing both the horizontal and the verticaledge information and pipeline operation in this thesis. Based on these characteristics,this thesis gives the architecture of hardware implementation, describes the functionof each module, hardware implementation and logic circuit diagram. Finally, thesimulation and verification of the design are given, including the functionalsimulation of each module and the system verification of FPGA. The results show thatthis design is available. |
