Equipment and super-resolution video scaling method robust

摘要

An apparatus and method for scaling the video image, it is a method and apparatus is applied to a technique of performing the de-interlacing converts the format of the video sequence and more specifically. The present invention includes a (101) interpolator, which is parameterized to calculate the interpolated value of each new point of the output. The calculated interval is estimated estimator insensitive type section (102), and has its maximum value is the probability distribution of the output image value, calculation of the insensitive type loss value (103), the section - from the insensitive type probability distribution measured value of the negative log-likelihood is determined for each interpolated value. After spatial normalization (104), the present invention selects the minimum loss interpolated value each new point, and a (105) provides minimum loss value selector as an output image value and this value. It is possible that further, the present invention also includes a directional matching error module and total variation incremental module and penalized the insensitive type loss value for the minimum loss value selected by this.