基于ASHE和SWT的低对比度自然场景图像文字定位  

TEXT LOCALIZATION OF LOW CONTRAST NATURAL SCENE IMAGES BASED ON ASHE AND SWT

作　　者：徐武[1] 杨昊东 陈盈君 汤弘毅 Xu Wu;Yang Haodong;Chen Yingjun;Tang Hongyi(Institute of Electrical and Information Engineering,Yunnan Minzu University,Kunming 650500,Yunnan,China)

机构地区：[1]云南民族大学电气信息工程学院,云南昆明650500

出　　处：《计算机应用与软件》2023年第3期229-234,共6页Computer Applications and Software

基　　金：国家自然科学基金项目(U1802271)。

摘　　要：为解决在自然场景下光照不均匀使得图像对比度低,导致图像文字定位错检的问题,采用自适应子直方图均衡算法(Adaptive Sub-Histogram Equalization,ASHE)对图像进行对比度增强处理;采用最大稳定极值区域算法提取候选区域,结合形态学与启发式规则进行初步定位;通过改进的笔画宽度变换算法(SWT)与汉字特征进行精确定位;使用字符链融合算法,将文字区域合并,实现最终文本定位。实验结果表明改进算法对自然场景低对比度图像中文字具有较高的定位准确率,并且有效地降低了时间复杂度。This paper aims to solve the problem that low contrast caused by uneven illumination in natural scene leads to false detection of text localization.The adaptive sub-histogram equalization algorithm(ASHE)was used to enhance the contrast of the image.The maximum stable extremum region algorithm was used to extract candidate regions,combined with morphology and heuristic rules for preliminary positioning.The improved stroke width transform algorithm(SWT)was used to accurately locate the Chinese character characteristics.The character chain fusion algorithm was used to merge the text regions to achieve the final text positioning.The experimental results show that the improved algorithm has higher localization accuracy for text in low contrast natural scene images,and effectively reduces the time complexity.

关 键 词：对比度 文字定位 自适应子直方图均衡算法 笔画宽度变换 

分 类 号：TP391.43[自动化与计算机技术—计算机应用技术]