Image Re-attentionizing Using Particle Swarm Optimization

In this paper, a method has been made to increase the capability of drawing attention for a target object. The forceful intervention in redirecting attention towards an intended object is known as image re-attentionizing. In this paper, it is achieved through changing the color values of target object without destroying naturalness of an image. This proposed method estimates the changed color values of the target object through maximization of difference in color (which is modulated by positional proximity) with other objects in an image. Large modification in the target object color values can disturb the naturalness of an image. Hence, this adds a constraint in the proposed optimization problem. An approach based on particle swarm optimization is proposed here to derive the new color values to optimize the saliency of the target object. Experimental results involving a saliency map-based objective evaluation and gaze-tracking experiments demonstrate the suitability of the proposed method.