计算机科学 ›› 2024, Vol. 51 ›› Issue (9): 147-154.doi: 10.11896/jsjkx.230800003
王笳辉, 彭光灵, 段亮, 袁国武, 岳昆
WANG Jiahui, PENG Guangling, DUAN Liang, YUAN Guowu, YUE Kun
摘要： 阴影消除是计算机视觉领域中面对阴影场景的重要任务,旨在检测和消除图像中的阴影区域。由于图像编辑技术受到阴影图像质量的制约,现有方法利用其他任务中的知识和阴影特性来获得更加有效的特征向量,从而实现阴影消除。在带有文本内容的阴影图像中,由于文本颜色和形状等特征不同于前景和背景,因此可能将文本错误地检测为阴影的一部分进而导致错误的阴影消除结果。针对该问题,提出了一种面向文本识别的小样本阴影消除方法。在小样本目标检测基础框架模型中,利用被错误识别为阴影的文本特征生成基类数据和新类数据,增强对该类文本的特征学习;在部分检测框合并算法中,利用文本本身长宽比多样化、变化大的特性,以多个约束为前提合并结构相关性较强的检测框,实现对目标的正确框定。建立在真实数据与合成数据上的实验结果验证了所提方法的有效性。
中图分类号:
| [1]BAKO S,DARABI S,SHECHTMAN E,et al.Removing sha-dows from images of documents [C]//The 13th Asian Confe-rence on Computer Vision.2016:173-183. [2]LIN Y H,CHEN W C,CHUANG Y.BEDSR-Net:A DeepShadow Removal Network from a Single Document Image [C]//IEEE Conference on Computer Vision and Pattern Recognition.2020:12902-12911. [3]EVERINGHAM M,VAN G,WILLIAMS C,et al.The PASCAL Visual Object Classes Challenge 2007(VOC2007) Results [EB/OL].http://www.pascal-network.org/challenges/VOC/voc2007/workshop/index.html.2007. [4]WANG Y P,ZHAO X,LI Y,el al.Densely cascaded shadow detection network via deeply supervised parallel fusion [C]//International Joint Conference on Artificial Intelligence.2018:1007-1013. [5]VICENTE T,HOU L,YU C,et al.Large-scale training of sha-dow detectors with noisily-annotated shadow examples [C]//European Conference on Computer Vision.2016:816-832. [6]HU X W,FU C W,ZHU L,et al.Direction-Aware Spatial Context Features for Shadow Detection [C]//IEEE Conference on Computer Vision and Pattern Recognition.2018:7454-7462. [7]WANG X,HUANG T,GONZALEZ J,et al.Frustratingly Sim-ple Few-Shot Object Detection [C]//International Conference on Machine Learning.2020:9919-9928. [8]YANG Q,TAN K,AHUJA N.Shadow removal using bilateral filtering [J].IEEE Transactions on Image Processing,2012,21(10):4361-4368. [9]ZHANG L,ANDY M,TAN C.Removing shading distortions in camera-based document images using inpainting and surface fitting with radial basis functions [C]//International Conference on Document Analysis and Recognition.2007:984-988. [10]TSOI Y,BROWN M.Geometric and Shading Correction forImages of Printed Materials a Unified Approach Using Boundary[C]//IEEE Conference on Computer Vision and Pattern Recognition,2004:240-246. [11]JUNG S,ABUL H,KIM C.Water-filling:An efficient algorithm for digitized document shadow removal [C]//Asian Conference on Computer Vision.2018:398-414. [12]KLIGLER N,KATZ S,AYELLET T.Document enhancement using visibility detection [C]//IEEE Conference on Computer Vision and Pattern Recognition.2018:2374-2382. [13]OLIVEIRA D,LINS R,GABRIEL F.Shading removal of illustrated documents [C]//International Conference on Image Analysis and Recognition.2013:308-317. [14]FINLAYSON D,HORDLEY S D,LU C,et al.On the removal of shadows from images [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2006,28(1):59-68. [15]TIAN J D,QI X J,QU L Q,et al.New spectrum ratio properties and features for shadow detection [J].Pattern Recognition,2016,51(3):85-96. [16]GUO R Q,DAI Q Y,HOIEM D.Single-image shadow detection and removal using paired regions [C]//IEEE Conference on Computer Vision and Pattern Recognition.2011:2033-2040. [17]HUANG X,HUA G,TUMBLIN J,et al.What characterizes ashadow boundary under the sun and sky? [C]//IEEE International Conference on Computer Vision.2011:898-905. [18]VICENTE Y,TOMAS F,HOAI M,et al.Leave-one-out kerneloptimization for shadow detection [C]//IEEE International Conference on Computer Vision.2015:3388-3396. [19]ZHU J J,SAMUEL K,MASOOD S,et al.Learning to recognize shadows in monochromatic natural images [C]//IEEE Confe-rence on Computer Vision and Pattern Recognition.2010:223-230. [20]KHAN S,BENNAMOUN M,SOHEL F,et al.Automatic feature learning for robust shadow detection[C]//IEEE Confe-rence on Computer Vision and Pattern Recognition.2014:1939-1946. [21]SHEN L,CHUA T,LEMAN K.Shadow optimization fromstructured deep edge detection [C]//IEEE Conference on Computer Vision and Pattern Recognition.2015:2067-2074. [22]VICENTE T,HOU L,YU C,et al.Large-scale training of sha-dow detectors with noisily-annotated shadow examples [C]//European Conference on Computer Vision.2016:816-832. [23]SHI H,ZHANG L.Image Shadow Removal Algorithm Based on Generative Adversarial Network [J].Computer Science,2021,48(6):145-152. [24]HU X W,FU C W,ZHU L,et al.Direction-aware spatial context features for shadow detection[C]//IEEE Conference on Computer Vision and Pattern Recognition.2018:7454-7462. [25]WANG J F,LI X,YANG J.Stacked conditional generative adversarial networks for jointly learning shadow detection and shadow removal [C]//IEEE Conference on Computer Vision and Pattern Recognition.2018:1788-1797. [26]ZHU Y,HUANG J,FU X,et al.Bijective mapping network for shadow removal [C]//IEEE/CVF Conference on Computer Vision and Pattern Recognition.2022:5627-5636. [27]ZHU Y,XIAO Z,FANG Y,et al.Efficient model-driven network for shadow removal [C]//AAAI Conference on Artificial Intelligence.2022:3635-3643. [28]LUO W,XIE X,Deng K,et al.Learning Shadow Removal from Unpaired Samples via Reciprocal Learning [J].IEEE Transactions on Image Processing,2023,32:3455-3464. [29]REN S,HE K,ROSS B,et al.Faster R-CNN:Towards Real-Time Object Detection with Region Proposal Networks [C]//Neural Information Processing Systems.2015:91-99. [30]HUANG W L,LIN Z,YANG J C,et al.Text localization in na-tural images using stroke feature transform and text covariance descriptors [C]//IEEE International Conference on Computer Vision,2013:1241-1248. |
|
||
首页