An Efficient Motion Adaptive De-interlacing and Its VLSI Architecture Design Hongbin Sun, Nanning Zheng, Chenyang Ge, Dong Wang and Pengju Ren The Institute of Artificial Intelligence and Robotics, Xi’an Jiaotong University No. 28, Xianning West Road, Xi’an, Shaanxi, 710049, P. R. China {hbsun, nnzheng, cyge, dwang, pjren}@aiar.xjtu.edu.cn Abstract This paper presents an efficient motion adaptive de- interlacing technique that consists of two main steps, i.e. 4-field extended Gaussian filtering motion detection and adjustable window ELA de-interlacing. Four consecutive interlaced fields are used to detect motion accurately. With a Gaussian filter, the motion detection can eliminate the influence of noise. And, an adjustable window ELA de-interlacing is adopted, which can reconstruct image with high quality even in areas with horizontal edge and texture. Experimental results show that the proposed algorithm outperforms previous methods both objectively and subjectively. Based on our method, the high speed VLSI architecture has been designed and implemented. The operating frequency of the chip is 126MHz that it could process format conversion in real-time for both SDTV and HDTV applications. 1. Introduction At present, there are two kinds of video format, i.e. interlace video and progressive video. With recent advances of HDTV and Multimedia Devices, we need much more progressive videos for our display devices such as LCD-TVs, Plasma Display Panels, etc. High quality de-interlacing becomes an important technology, which converts original interlaced video sequences into progressive frames without artifacts. In the past few decades, many methods for de- interlacing have been proposed. These methods can be roughly categorized as non-motion compensated and motion compensated ones [1]. Although motion compensated methods produce the best reconstructed quality, they are too computationally expensive for real-time video applications. Therefore, we should introduce non-motion compensated methods, especially for real-time applications such as video compression and digital TV. Various non-motion compensated de-interlacing methods have been proposed. These methods include temporal, spatial and motion-adaptive schemes. The simplest temporal method is directly combining two interlaced fields into one progressive frame. However, the line-crawling effect will occur in the motion area. The edge-based line average (ELA) [2], [3] is a spatial method widely used since it exhibits fairly good performance with simple calculation. However, the shortcoming of ELA-based methods is that they may deteriorate the picture quality in static area and are sensitive to the high frequency area. To solve the aforementioned problems, motion adaptive techniques have been developed. Motion adaptive de-interlacing methods separate the picture with a motion detector first, then adopt temporal filter in the static area and spatial filter in the motion area. Motion adaptive de-interlacing methods can provide high-resolution and flick-free pictures with acceptable computational complexity. Therefore, they are widely used in real-time video applications. Motion adaptive methods are dependent on accurate motion detection, any false detection can conclude visible artifact. Vertical temporal (VT) median filtering is an implicitly adaptive method, which can adapt to different area without motion detection. In this paper, a motion adaptive de-interlacing method based on adjustable window ELA (AW-ELA) with accurate motion detection is proposed. The remainder of this paper is organized as follows. In section 2, the proposed motion adaptive de-interlacing method is detailed. Section 3 shows the experimental results and comparison with previous methods. Section 4 aims to describe the architecture design of proposed method in ASIC hardware. Section 5 gives the conclusion and remarks.