Panoramic stitched image quality evaluation method based on deep learning

• Main Authors: TIAN CHONGZHEN, SHAO FENG
• Format: Patent
• Language: Chinese; English
• Published: 03.02.2023
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; PHYSICS

The invention discloses a panoramic spliced image quality evaluation method based on deep learning, and the method comprises the steps: carrying out the viewport extraction of a distorted and reference panoramic spliced image in a training stage, obtaining a distorted and reference viewport image, inputting the distorted and reference viewport image into a distortion correction network for training, outputting a pseudo reference viewport image, and carrying out the reconstruction of the pseudo reference viewport image. Training to obtain a distortion correction network training model; inputting the distortion, the reference viewport image and the average subjective score value into a quality evaluation network for training, outputting a quality prediction score, and training to obtain a quality evaluation network training model; the distortion correction network training model performs distortion correction on the distorted viewport image in a test stage to generate a pseudo reference viewport image, and fina