Image Saliency Detection Based on Contrast Feature of Superpixel Patch
Chao Jia, Hailan Yu, Qilong Chen · 2020
Saliency detection based on visual attention mechanism is one of the current researches focuses of computer vision. In this paper, an image saliency detection algorithm based on the contrast feature of superpixel patch is proposed. First, RGB color space image is transformed into HSV color space, and color saliency is calculated. Then, the SLI C algorithm is used to construct the superpixel patch structure of the image. In LAB color space, the contrast saliency is calculated. Finally, the salient images of color feature and contrast feature are fused and multi-scale enhanced. Image saliency detection is carried out on MSRA - 1000 data set. Compared with the existing methods, the algorithm in this paper detects more complete contour of image saliency region, has better precision and recall rate, and can obviously suppress complex texture and noise.