本研究透過客觀的視訊品質量測法分析不同特徵的視訊內容,建立視訊類型的量化分類依據,使用尖峰訊號雜訊比(Peak Signal Noise Ratio, PSNR)瞭解影響變數之間的關連性,同時也利用主觀的視訊品質量測方法,來瞭解觀眾對於不同的視訊內容特徵、不同的畫面播放率與不同的資料傳輸率對視訊品質的影響。主觀視訊品質評估的實驗設計是採用雙受測連續性品質評分法(Double Stimulus Continuous Quality Scale, DSCQS)。 在我們的實驗裡,讓受測者針對參考視訊序列與測試視訊序列進行感受滿意度之評分,並將本實驗的主觀品質評估資料以統計分析找出不同內容特徵的影響。研究成果發現受測者觀察低移動特徵的視訊內容時,均偏好低畫面播放率,而觀察高移動特徵時,則偏好較高的畫面播放率,另外,我們也透過客觀視訊特徵分類方法發現不同的視訊內容在主觀品質評估上,其不同的資料傳輸率皆有顯著差異,而畫面播放率對不同視訊內容則為部份差異。 我們利用LISREL等統計方法進行實證分析,以瞭解不同視訊內容特徵的影響性,例如低移動與簡單紋理特徵、高移動與簡單紋理特徵、低移動與複雜紋理特徵及高移動與複雜紋理特徵之序列,由分析結果發現畫面播放率與資料傳輸率對於不同視訊內容之畫面解析度、畫面流暢度與整體滿意度有著不同層級的影響。
This paper describes different visual perception for the subjects in different frame rates, bit rates ,and video contents. Our analysis for different video contents uses the objective video quality evaluation by the Peak Signal Noise Ratio (PSNR). At the same time, this paper performed an experiment for subjective quality evaluation employs the Double Stimulus Continuous Quality Scale (DSCQS) methodology. In our experiment, the subjects assess the scores of perceived quality by comparing the reference sequences with the test sequences. In the experimental results, we find that the video contents with low motion characteristic prefer low frame rate and the sequences with high motion characteristic prefer high frame rate. In addition, the perceived quality for different content characteristic has significant and partial significant differences under different bit rates and frame rates, respectively. In the empirical study, we analyze also the mean opinion scores of the subjects by the statistic tool, LISREL, to understand the influence levels of different content characteristics such as the low motion and easy texture, the high motion and easy texture, the low motion and detailed texture, and the high motion and detailed texture sequences. The results and findings indicate that the frame rate and the bit rate influence the different levels on the spatial sharpness, the temporal continuity, and the integral satisfaction.