METHOD FOR MEASURING VIDEO QUALITY USING MACHINE LEARNING BASED FEATURES AND KNOWLEDGE BASED FEATURES AND APPARATUS USING THE SAME

摘要

Disclosed are a method for automatically measuring video quality using machine learning-based features and knowledge-based features, and an apparatus therefor. In the method for automatically measuring video quality according to an embodiment of the present invention, a target video for measuring quality, a reference video to be compared, and a knowledge-based feature are input into a machine learning-based frame-by-frame feature extraction model to obtain a frame-by-frame quality prediction value. Is calculated, and the frame score vector is generated by combining all the predicted quality values of each frame, and the frame score vector is inputted to the convolutional neural network to calculate a temporal weight having the same size as the frame score vector, and the frame score vector and the temporal weight The result of the dot product is input to the fully connected neural network, and the predicted quality of the target video is output.