| 摘要 |
<p>A learning apparatus and a data converting apparatus for allowing an image to be converted to an image whose quality has been improved. A class-tap producing circuit (2) and a prediction-tap producing circuit (3) produce, from an SD image, a prediction-tap for use in determining the HD pixels of a notice block of an HD image. A classifying circuit (4) classifies, based on the class-tap, the HD pixels of the notice block. A coefficient RAM (7) acquires the tap coefficient of the class of the HD pixels of the notice block from the tap coefficients obtained by applying a predetermined constraint condition to the training data to learn the relationship between the training and learning data every at least one class. A prediction arithmetic circuit (8) and a decoding circuit (9) use the acquired tap coefficient and prediction tap to determine the HD pixels of the notice block.</p> |
