Modelling The Effects Of Focus On Target Saliency: A Texture And Edge-Based Method

No Thumbnail Available

Links to Files

Author/Creator ORCID

Date

2014

Department

Electrical and Computer Engineering

Program

Doctor of Engineering

Citation of Original Publication

Rights

This item is made available by Morgan State University for personal, educational, and research purposes in accordance with Title 17 of the U.S. Copyright Law. Other uses may require permission from the copyright owner.

Abstract

Visual search involves attending to various locations in a scene depending on two main factors: "bottom-up" processing or "top-down" processing. Bottom-up attention is attracted by objects that "pop-out" or stand-out from their background. Top-down attention is a guided process that focuses on objects in a scene based on some prior knowledge or task. Many times completing a search task involves the combination of top-down and bottom-up processing. In this situation, visual attention is directed towards objects that pop-out before the subject of the search task. During visual search, the human visual system is able to transition the saccade pattern of the eye to a more focused region of the attended scene. The benefit is the attended area is in detail versus the periphery. For example, when law enforcement personnel conduct visual search using a night-vision system, the device is set for a focus distance to attend to one location causing other areas of the scene to appear blurry. The depth of focus of the device establishes how much of the scene is in focus. The depth of focus of a night vision device may affect how some objects in the scene are able to stand-out from their surroundings, the saliency of the object. In this research we concentrate on the creation of a saliency model based on texture and edge density. This model evaluates the saliency of targets in a scene with respect to the focus range used to capture the image. The dataset included static images with different focal conditions. The four focus conditions used for the experiment included a near focus ranging between 1 ft to 10 ft, a midrange focus ranging between 10 ft to 50 ft, a midrange focus ranging between 10 ft and 50ft with a reduced aperture and a far focus ranging between 50ft to infinity. To validate the model, a human factors study was completed to evaluate target detection and reaction times with the four focal conditions. Both research efforts concluded that the focal condition with a reduced aperture presented the best depth of focus within the scene. The output of the model showed a higher target saliency with the reduced aperture condition than with the other three focal conditions. The human performance data validated that information with participants detecting and reacting faster to targets with the reduced focus condition. Participants in the reduced aperture condition detected an average of 5.19% more targets than the participants in the other three focal conditions. Participants in the reduced aperture condition reacted to targets an average of 10.33% faster than the participants in the other three focal conditions.