Future image frame generation using Artificial Neural Network with selected features

This paper presents a novel approach for the generation of Future image frames using Artificial Neural Network (ANN) on spatiotemporal framework. The input to this network are hyper-dimensional color and spatiotemporal features of every pixel of an image in an image sequence. Principal Component Analysis, Mutual Information, Interaction Information and Bhattacharyya Distance measure based feature selection techniques have been used to reduce the dimensionality of the feature set. The pixel values of an image frame are predicted using a simple ANN back propagation algorithm. The ANN network is trained for R, G and B values for each and every pixel in an image frame. The resulting model is successfully applied on an image sequence of a landing fighter plane. As Mentioned above four feature selection techniques are used to compare the performance of the proposed ANN model. The quality of the generated future image frames is assessed using, Canny edge detection based Image Comparison Metric(CIM) and Mean Structural Similarity Index Measure(MSSIM) image quality measures. The proposed approach is found to have generated six future image frames successfully with acceptable quality of images.