pyramid coding is one of the most promising schemes for multi-resolutional representation (MRR) of images. However the number of Pixels of pyramid schme is larger than that of the original image. Previously, we proposed complementary pyramid which has no redundant pixels. In this paper, we analyze theoretically the performance of complementary pyramid as MRR and compare it with the performance of other MRR coding schemes. Coding gain of complementary pyramid is nearly equal to that of wavelet transform and subband scheme with SSKF, and higher than that of conventional pyramid. Better multi-resolutional images are achieved in the complementary pyramid because of its liner phase characteristic.

View full abstract

This paper describes analysis of a system which transmits multiplexed NTSC TV signals for broadcasting satellite. This system compresses TV signals by TAT, and transmits the multiplexed signals by Differential FM. System parameters such as number of multiplexed signals. SN ratio and waveform distortion are analized. It is shown that 4 TV programs can be transmitted over a single satellite channel with good SN ratio and waveform distortion.

View full abstract

This paper describes a new chrominance signal processing method for moving areas of MUSE system. It produces an improvement in picture quality. This is realized by applying intra-frame signal processing to moving areas. Computer simulation on moving pictures have produced a good resulut. It has compatibility with the current MUSE system and is easy to adopt it for the MUSE encoder and decoder.

View full abstract

TV電話への適用を目的とした,人物抽出処理について検討した.ここで提案する方法は2フレ-ム間の輝度信号を差分処理して,切れ目のない領域輪郭を形成した後,3×3の画素サイズのウインドウ内において領域塗りつぶし処理とウインドウ移動位置の探索を行う.ウインドウは4方向に移動可能な折り畳み走査経路に沿って移動させ,効率的な領域抽出を行う.本処理法は連続フレ-ム間からの動領域抽出および背景との差分からの動領域抽出のいずれにも適用でき,処理結果は良好な人物抽出を実現している.

View full abstract

Aiming at realizing a super high efficiency video coding for communication, we are studying the media conversion method which creates lip movements of talking head images from the text information. In this paper, we focused on the implementation of this method onto the PC-based real time system. Reflecting on the restricted computing ability of the PC, we simplified the algorithm implimented. As the result, we could realize the experimental real time system composed of the tiny software on the PC with the commercially available voice synthesizing unit, which generates the talking head images and voices according to the input of the text data.

View full abstract