Image Retargeting Using Depth Enhanced Saliency

Most of existing image retargeting approaches only use color information to help preserve important image content. In this paper, we exploit both color and depth information captured by an RGB-D camera to preserve important regions and maintain background structure. In particular, we introduce a depth-based importance map, such that deformation of the image is guided by this map. This depth-based importance map is produced automatically by combining gradient, salience, and depth-based measures. The depthbased importance map is subsequently utilized as the input to the discontinuous seam carving algorithm, and the seam is processed by the dynamic programming method. Our experimental results demonstrate the effectiveness of using depth-based importance map with the discontinuous seam carving algorithm. The proposed method performs better in terms of preserving important regions and maintaining image background structure. Keywordscontent-aware image retargeting; visual attention; depth informatio; RGB-D camera.