基于场景分析的建筑图像水平自适应调整方法  

Architecture image horizontal adaptive adjustment based on scenario analysis

作　　者：王嶺 许颖浩 沈笠[3] 周军[3] 

机构地区：[1]上海文广科技(集团)有限公司,上海200233 [2]上海文广互动电视有限公司,上海200072 [3]上海交通大学图像通信与网络工程研究所,上海200240

出　　处：《电视技术》2016年第6期5-9,共5页Video Engineering

基　　金：国家科技支撑计划项目(2015BAK05B03);上海张江国家自主创新示范区专项发展项目(201411-JA-B108-002)

摘　　要：建筑图像是一类经常会在电视节目中出现的内容。摄像机非水平位置拍摄建筑物会影响建筑图像画面的美感。建筑类自然场景图像倾斜自动调整,其技术核心在于如何检测出图像拍摄时的水平倾斜角。针对建筑图像的自适应水平调整要求,基于成角透视原理,利用建筑的轮廓线,通过边缘检测及霍夫变换提取出主要偏向水平直线的参数,基于直线参数方程得到平行线的透视消失点,再利用建筑物的两个水平消失点估计得到成像倾角,最后通过图像旋转变换得到调整后的图像,相关实验结果验证了所提方法的有效性。Architectural is a kind content that often occurs in various types of TV programs. The camera placed in a non-horizontal position will affect the beauty of the captured architectural image. The core technology for inclined architectural image adjustment is how to detect the level of the inclined shooting angle. In this paper, an automatic inclined architecture image adjustment method is proposed based on the principle of angled perspective. The edge detection and Hough transform are applied to extract the architec- ture＇ s contour. And then two vanishing points for perspective parallel lines are estimated to get the inclined angle of the image for image horizontal adjustment. The experiment results demonstrate the method proposed in this paper.

关 键 词：边缘检测 霍夫变换 图像水平 

分 类 号：TN941.1[电子电信—信号与信息处理]