复杂背景下视觉传达图像目标提取方法  

Method for Targets Extraction of Visual Communication Image in Complex Backgrounds

作　　者：陈宁 CHEN Ning(Department of Fine Arts,Bozhou University,Bozhou 236800,China)

机构地区：[1]亳州学院美术系,安徽亳州236800

出　　处：《许昌学院学报》2025年第2期106-110,共5页Journal of Xuchang University

基　　金：安徽省高校哲学社会科学研究重点项目(2023AH052254)。

摘　　要：为保证视觉传达图像目标提取质量,提出复杂背景下视觉传达图像目标提取方法.采用视觉传达技术增强图像,进行阈值分割获取感兴趣区域,识别关键特征.结合局部直方图,用模糊加权算子平抑杂散效应,得到显著性矩阵,并求边缘检测算子.将距离质心近的像素点作为轮廓点,线性连接得图像轮廓,实现复杂背景下目标提取.实验结果表明,利用新方法提取图像目标的结构相似度均保持在0.87以上,提取质量较高.In the extraction of image target elements,due to the influence of complex backgrounds and spurious effects,the structural similarity between the extracted image target and the original image is low,and the extraction quality cannot be guaranteed.Therefore,a method for extracting visual communication image targets in complex backgrounds is proposed.Visual communication technology is used to perform panoramic reconstruction and color mode transformation on the original image,extracting high-frequency and low-frequency components.The original image is enhanced by combining the image spatial domain function,and the enhanced image is subjected to threshold segmentation to obtain the interest region of the image,in order to identify the key features of the image.Combined with the local histogram of the image,a fuzzy weighting operator is introduced to smooth out the spurious effect,thereby obtaining the saliency matrix of the image.The edge detection operator of the image pixels is obtained,and the pixels with a distance less than the centroid Mahalanobis distance are used as contour points.Linear connections are used to detect the shape contour of the image,thus achieving the extraction of image targets in complex backgrounds.The experimental results show that using the proposed method to extract image targets,the structural similarity obtained remains above 0.85,indicating high extraction quality.

关 键 词：复杂背景 视觉传达 感兴趣区域 特征识别 元素提取 图像质量 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]