Abstract:
Saliency is the quality of an object that makes it stands out from neighboring items and grabs viewer attention. In regards to image processing, it refers to the pixel or group of pixels that stand out in an image or a video clip and capture the attention of the viewer. Our eye movement is usually guided by saliency while inspecting a scene.
Rapid detection of emotive stimuli is the ability of humans. Visual objects in the scene are also emotionally salient. As different images and clips can cause different emotional responses in the viewer such as happy or sad, therefore, there is a need to measure these emotions along with the visual saliency.
This study is conducted in order to determine if the existing visual saliency models also measure the emotional saliency. The model used in the study is Graph Based Visual Saliency (GBVS) model. This model assigns higher saliency value to the center of an image. According to the experiments conducted on videos from publically available Antoine Coutrot database, the result shows that there is low saliency or salient features in sad movies hence making these videos less emotionally salient. However, over all visual content does not capture emotional salience. The Graph based visual Saliency Map developed failed to capture the sad emotion. Therefore, there is a requirement that such classical model to be developed which can capture emotional content.
