This paper presents a novel method for salient object detection that improves upon existing approaches by introducing a robust background measure called boundary connectivity and a principled optimization framework for integrating multiple saliency cues. The proposed boundary connectivity measure quantifies how connected a region is to image boundaries, providing a more reliable and stable background detection than previous methods. It is defined as the ratio of a region's perimeter on the boundary to its overall perimeter, or the square root of its area, and is robust to variations in image content. This measure is used to enhance traditional contrast-based saliency computation, leading to more accurate and uniform saliency maps. The paper also introduces a principled optimization framework that integrates multiple low-level cues, including the new background measure, to achieve state-of-the-art results. The framework formulates the saliency estimation as a global optimization problem, with a cost function that enforces constraints on object and background regions, ensuring smoothness in the saliency map. The optimal saliency map is obtained by solving this cost function using efficient least-square optimization. Experiments on several benchmark datasets show that the proposed method outperforms existing approaches in terms of accuracy and robustness. The method is particularly effective in handling complex backgrounds and objects that are only slightly touching the image boundaries. The results demonstrate that the proposed background measure and optimization framework significantly improve the performance of salient object detection, making it more reliable and efficient. The method is also efficient, with a running time of less than 3 milliseconds for 200 superpixels, making it suitable for real-time applications.This paper presents a novel method for salient object detection that improves upon existing approaches by introducing a robust background measure called boundary connectivity and a principled optimization framework for integrating multiple saliency cues. The proposed boundary connectivity measure quantifies how connected a region is to image boundaries, providing a more reliable and stable background detection than previous methods. It is defined as the ratio of a region's perimeter on the boundary to its overall perimeter, or the square root of its area, and is robust to variations in image content. This measure is used to enhance traditional contrast-based saliency computation, leading to more accurate and uniform saliency maps. The paper also introduces a principled optimization framework that integrates multiple low-level cues, including the new background measure, to achieve state-of-the-art results. The framework formulates the saliency estimation as a global optimization problem, with a cost function that enforces constraints on object and background regions, ensuring smoothness in the saliency map. The optimal saliency map is obtained by solving this cost function using efficient least-square optimization. Experiments on several benchmark datasets show that the proposed method outperforms existing approaches in terms of accuracy and robustness. The method is particularly effective in handling complex backgrounds and objects that are only slightly touching the image boundaries. The results demonstrate that the proposed background measure and optimization framework significantly improve the performance of salient object detection, making it more reliable and efficient. The method is also efficient, with a running time of less than 3 milliseconds for 200 superpixels, making it suitable for real-time applications.