XU Xiuni,HU Ziwei.Research on enhancing the resolution of reconstructed single frame character image[J].Journal of Yanbian University,2020,46(03):221-225.
增强重建单帧字符图像分辨率的方法研究
- Title:
- Research on enhancing the resolution of reconstructed single frame character image
- 文章编号:
- 1004-4353(2020)03-0221-05
- 分类号:
- TP391
- 文献标志码:
- A
- 摘要:
- 针对传统单帧字符图像重建方法存在分辨率较低的问题,提出了一种增强重建单帧字符图像分辨率的方法.首先,构建单帧字符图像重建的三维轮廓特征检测模型; 其次,采用边缘轮廓分割法提取单帧字符图像的三维细节特征,并结合小波多尺度分解法实现尺度特征分离; 然后,采用小波阈值去噪法对图像进行滤波降噪,并使用非负邻域嵌入的方法增强小波阈值辨识度,以此构建增强重建单帧字符图像分辨率的模型,完成单帧字符图像的高分辨率重建.最后,以Matlab 2013软件为实验平台进行仿真测试显示,采用本文方法得到的重建图像的清晰度明显优于文献[2-4]方法,且平均峰值信噪比(26 dB)也显著优于文献[2](15 dB)、文献[3](10 dB)、文献[4](16 dB)的方法.因此,本文方法在车牌号码识别、基于内容的图像检索、文档图像分析等领域中具有较好的应用价值.
- Abstract:
- To solve the problem of low resolution in traditional single frame character image reconstruction method, a single -frame character image reconstruction resolution enhancement method is proposed. Firstly, a three -dimensional contour feature detection model for single -frame character image reconstruction is formed. Second, the edge contour segmentation method is used to extract the three -dimensional detailed features of a single frame character image, and the wavelet multi -scale decomposition method is used to achieve the separation of scale features. Then wavelet threshold denoising method is used to denoise the image. The conversion noise reduction is performed, and the non -negative neighborhood embedding method is used to enhance the wavelet threshold recognition. Finally, Matlab 2013 software is used as the experimental platform for simulation test, the results show that the resolution of the reconstructed image obtained by this method is better than that of the references [2-4], and the average peak signal -to -noise ratio(26 dB)is better than that of literature [2](15 dB), literature [3](10 dB), and literature [4](16 dB). This method has good application value in applications such as license plate number recognition, content -based image retrieval, and document image analysis.
参考文献/References:
[1] 魏光杏,周献中,卜锡滨.基于模糊C均值与人工蜂群优化的灰度图像分割[J].兰州大学学报(自然科学版),2019,55(2):250-254.
[2] 李键红,吴亚榕,吕巨建.基于自相似性与多任务高斯过程回归的单帧图像超分辨率重建[J].光学精密工程,2018,26(11):2814-2826.
[3] 赵丽玲,孙权森,张泽林.基于深度学习特征字典的单帧图像超分辨率重建[J].数据采集与处理,2018,33(4):740-750.
[4] 苏健民,杨岚心.基于生成对抗网络的单帧遥感图像超分辨率[J].计算机工程与应用,2019,55(12):202-207.
[5] 余玛俐,张海.基于低秩矩阵填充与全变分约束的HDR成像[J].计算机工程,2019,45(4):262-266.
[6] GRAF B L, ROJO L E, DELATORRE -HERRERA J. ChromoTrace: Reconstruction of 3D chromosome configurations by super -resolution microscopy[J]. Food Chemistry, 2017,131(2):387-396.
[7] MOON H M, CHANG H S, PAN S B. A low -cost media quality enhancement resolution up -conversion for mobile cloud[J]. Journal of Supercomputing, 2017,73(7):3098-3111.
[8] WANG Y L, ZHOU Y N, WANG S X, et al. Enhancement of spatial resolution of ghost imaging via localizing and thresholding[J]. Chinese Physics B, 2019,28(4):194-199.
备注/Memo
收稿日期: 2019-11-08
作者简介: 徐秀妮(1980—),女,讲师,研究方向为智能测控系统、交通信息工程及控制.