Image completion using prediction concept via support vector regression

Image completion is a widely used method for automatically removing objects or repairing the damaged portions of an image. However, information of the original image is often lacking in reconstructed structures; therefore, images with complex structures are difficult to restore. This study proposes a prediction-oriented image completion mechanism (PICM), which applies the prediction concept to image completion using numerous techniques and methods. The experiment results indicate that under normal circumstances, our PICM not only produces good inpainting quality but it is also easy to use.