基于显著性检测的全变差去高光研究  

Research on Total Variation De-Highlighting Based on Saliency Detection

作　　者：李佳俊 皮大能 代灿威 陈强 LI Jia-jun;PI Da-neng;DAI Can-wei;CHEN Qiang(School of Electrical Engineering and Automation,Hubei Normal University,Hubei Huangshi 435002,China)

机构地区：[1]湖北师范大学电气工程与自动化学院,湖北黄石435002

出　　处：《计算机仿真》2024年第3期214-218,共5页Computer Simulation

基　　金：国家自然科学基金资助项目(52075063);国家自然科学基金(61905172);山西省重点研发计划项目(202102020101005);山西省科技成果转化引导专项资助(202204021301059)。

摘　　要：选矿浮选过程中浮选槽中的泡沫图像,受到工业摄像角度和光照点位置影响,导致泡沫图像颜色特征以及纹理特征的提取达不到预期效果。为解决上述问题,提出一种基于显著性检测的自适应全变差去高光算法。将处于RGB颜色空间的图像转换到处于YUV颜色空间中,根据显著值的大小,判定某个像素点是否为高光像素点,并修复图像高光区域。构建改进的全变差修复模型,并对图像修复模型完成求解。实验结果可知,所提算法对图像高光区域的识别以及细微处的处理具有一定的提高,且均方误差值与峰值信噪比在一定程度上得到了改善,可以有效的提取泡沫图像存在的亮点区域且修复。The froth image in the flotation cell during beneficiation flotaion is affected by the angle of industrial camera and the position of illumination,which leads to the extraction of color features and texture features of the foam image failing to achieve the desired results.To address the aforementioned issues,an adaptive algorithm of total variation highlight removal based on saliency detection was proposed.At first,the image in RGB color space was converted to YUV space,thus determining whether a pixel was highlighted according to the size of significance value.Meanwhile,the highlight area of image was repaired.Moreover,an improved model of total variation restoration was constructed,and then the model was solved.Experimental results show that the proposed algorithm has a certain improvement in recognizing image highlights and processing fine spots.Meanwhile,the mean square error and peak signal to noise ratio have been improved to a certain extent.This method also can extract and repair the bright spots in froth images effectively.

关 键 词：工业选矿 显著性检测 自适应 全变差 泡沫图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]