Real-time face-priority auto-focus and adaptive auto-exposure for digital and cell-phone cameras

Auto-focus (AF) and Auto Exposure (AE) are key features in consumer level digital and cell-phone cameras allowing users to focus automatically and set the exposure automatically. AF provides automatic focusing in order to get sharp images with no user intervention. AE is also a key factor for automatically obtaining images with the right amount of brightness exposure. While there are many algorithms in the literature addressing the issues of AF and AE, most of these algorithms do not take into consideration speed, accuracy and cost of deployment on actual digital and cell-phone camera platforms. The first part of this dissertation involves the development and implementation of a Face-Priority AF algorithm due to the fact that the great majority of pictures taken by consumers are of human faces. In Face-priority AF, the focusing decision is made based on a detected face area in the image, thus capturing a sharp picture of the face. Although a number of face detection algorithms have appeared in the literature, their use is limited when considering their real-time software deployment on resource limited digital or cell-phone camera processors. A fast software Face-priority AF algorithm is introduced in this dissertation by combining a skin color model with a computationally efficient shape processing scheme. This algorithm is compared with a conventional algorithm in terms of speed and accuracy. A real-time implementation of this Face-priority AF is achieved on an actual digital camera platform. The second part of this dissertation involves the development and implementation of a novel AE algorithm named Adaptive AE. It is shown that this algorithm provides a more effective auto-exposure solution, in particular in poor lighting conditions. Conventional AE methods utilize the image brightness in one way or another to set the right exposure level, whereas the developed AE method uses the maximization of an information theory measure. It is shown that the developed AE solution generates more effective exposure setting in terms of contrast and brightness deviation. A real-time implementation of the Adaptive AE algorithm is also achieved on an actual digital camera platform.