ABSTRACT

An original approach is proposed. Called the Pre-Segmented Region of Interest Classification Scene System (PSROI), this system is able both to classify a scene during a digital camera's pre-capture phase, and to determine image processor parameters that will determine the quality of the final picture: white balancing, exposure, and focus. Additionally, it also integrates with the focusing system. An image in the classification scene system is given a value, called the focus weight. The focus weight ensures that the selected classification scene is more suitable for user vision. In the scene classification system, three scenes are set: portrait, landscape, and beach/snow. In order to perform the classification more quickly, the operation region of the image shrinks, and less computation is applied to build the system. Experimental results show that the proposed system is capable of effectively classifying scenes into the right categories within 0.2 s..