【摘要】：提出一種基于詞袋模型的新的顯著性目標(biāo)檢測(cè)方法.該方法首先利用目標(biāo)性計(jì)算先驗(yàn)概率顯著圖,然后在圖像的超像素區(qū)域內(nèi)建立詞袋模型,并基于此特征計(jì)算條件概率顯著圖,最后根據(jù)貝葉斯推斷將先驗(yàn)概率和條件概率顯著圖進(jìn)行合成.在ASD、SED以及SOD顯著性目標(biāo)公開數(shù)據(jù)庫(kù)上與目前16種主流方法進(jìn)行對(duì)比,實(shí)驗(yàn)結(jié)果表明本文方法具有更高的精度和更好的查全率,能夠一致高亮地凸顯圖像中的顯著性目標(biāo).
[Abstract]:A new significant target detection method based on word bag model is proposed. In this method, the priori probabilistic salience map is first calculated by objectiveness, then the word bag model is established in the super-pixel region of the image, and the conditional probabilistic significant map is calculated based on this feature. A priori probability and a conditional probabilistic salience diagram are combined according to Bayesian inference. The experimental results show that the proposed method has higher precision and better recall, and can consistently highlight the salient targets in the image by comparing with 16 popular methods in the open database of ASD,SED and SOD significant targets.
【作者單位】： 南通大學(xué)電氣工程學(xué)院;南京理工大學(xué)計(jì)算機(jī)科學(xué)與工程學(xué)院;
【基金】：國(guó)家自然科學(xué)基金(61272220)資助~~
【分類號(hào)】：TP391.41

【相似文獻(xiàn)】

相關(guān)期刊論文 前1條

1 徐丹;唐振民;徐威;;融合顏色屬性和空間信息的顯著性物體檢測(cè)[J];中國(guó)圖象圖形學(xué)報(bào);2014年04期

，

本文編號(hào)：2375525