壓縮 MPEG-2 視訊頻寬之轉碼器的設計與實作

Abstract

在低頻寬、影像品質要求不高的視訊應用中，我們常需降低視訊訊號傳輸 時所需的頻寬，藉此能在有限的頻寬中傳送更多的視訊資料。這種降低視 訊訊號頻寬的過程我們稱之為轉碼。在本篇論文中，我們所設計的轉碼器 是藉由控制量化的精細程度來放大視訊訊號壓縮的比率，達到降低頻寬的 目的。視訊經過轉碼的過程後，影像品質會因為頻寬的降低而變差，如何 在給定的轉碼頻寬下提高影像的品質也是我們的轉碼器設計的重點之一。 根據視覺感官對複雜的影像畫面有較大的影像誤差容忍能力的特性，我們 依照影像的複雜度給予精細不同的量化處理，這就是適應式量化的過程。 在實驗結果中，我們所設計的轉碼器能夠有效地提昇壓縮視訊到我們所指 定的頻寬。在影像品質方面，適應式量化的過程在主觀上有達到提影像品 質的目的，但因為適應式量化無法減少平均的影像誤差，所以在客觀的影 像品質估計中不能看出改善的效果。 Recently, digital compressed video widely used in the boardcast media. These video signals always have a higher bit ratefor maintaining a higher image quality. In some application we shouldconverte video to the lower bit rate formats. The process is called transcoding.MPEG-2 is one of compressed vidoe standards. The bandwidth of MPEG-2 compressed video is due to the compression ratio when encoding that video.The difference of a transcoder and a encoder is that the input signal to thetranscoder is a compressed video signal signal. By analyzing the input compressed video, the transcoder decode the input compressed video partially,and convert it into a lower bit rate compressed signal. In this thesis, we discuss how to convert MPEG-2 video signal of a given bit rate into a lower bit rate format by the adjustment of quantization scale. When transcoding a compressed video, the image quality will be loss. In this thesis, we also discuss how to improve the image quality at the transcoding process.In the experiment results, it will be shown that the transcoder which we design can reduce the bandwidth of MPEG-2 compressed video to a given lower bit rate. We have also found significant improvements in image quality in subjective terms. Unfortunately, objective measures of image quality fail to bear this improvement in a convincing manner.