| Summary: | 碩士 === 國立虎尾科技大學 === 電機工程系碩士班 === 107 === In this paper, we proposed an algorithm based on color distribution and color contrast image saliency detection. The method firstly reduces the total number of colors in the image through color quantization, extracts saliency maps based on color and spatial position; and then separately extracts color contrast map saliency maps based on color histogram and saliency maps based on color distribution; Then merge the saliency maps obtained in the two stages to get a coarse saliency map.; Finally, the coarse saliency map is selected to get the saliency value and the influence of the spatial weight and the closed operation are added to obtain the final saliency map.
After using the public database to compare with other algorithms, the experimental results show that under the comprehensive evaluation of accuracy and recall rate, the proposed algorithm wins most algorithms. And some algorithms were preferable in a comprehensive assessment, but in terms of processing time, they require the extra time of several times compared to this paper.
|
|---|
