A low computational complexity algorithm for real-time salient object detection

Image saliency detection is a process for highlighting the most salient object in an image and presenting the image saliency map. The content of an image is chaotic, including a complex background, low contrast, and an irregular salient object appearance. To overcome these problems, many algorithms have high computational complexity. In this paper, an efficient and fast-performing saliency detection algorithm is proposed, which consists of initiation saliency map generation and saliency map refinement. In the generation stage, the color-based contrast prior and color-based spatial distribution prior are effectively described in the image. Subsequently, two prior results (contrast value and distribution value) are fused to obtain an initial saliency map. In the refinement stage, the initial saliency map is refined by visual focus and an adaptive salient object mask (SOM). Due to the simplicity of the proposed algorithm, the system can detect salient objects in real time. Experimental evaluation on the benchmark shows that the proposed method can achieve sufficient accuracy and reliability while showing the lowest execution time. Compared with other methods, the execution time of the proposed method can achieve 137 frames per second (FPS) for the dataset with average image size 386 x 292.

Automated summary

This study proposes an efficient and fast-performing image saliency detection algorithm, which consists of initiation saliency map generation and saliency map refinement. Experimental evaluation shows that the proposed method achieves sufficient accuracy and reliability while having the lowest execution time.