Abstract
Document layout analysis is a critical step in optical character recognition. Traditional handcraft feature-based methods cannot handle various formats to obtain high accuracy. Although, deep-learning based methods obtain satisfactory accuracy, they are not memory-efficient for low-memory devices such as mobile phone. To alleviate such problems, a memory-efficient approach to layout analysis with the Lightweight Dilated Network (LD-Net) is proposed in this study. The initial document page image is segmented into blocks of content via Otsu algorithm and RLSA. Each block is sent into the LD-Net to classify them into four common different classes, figure, table, text, and formula. The main structure of the LD-Net is a shallow network, which performs better than deeper network for layout analysis task. Each convolution layer is composed of depthwise separable convolution and residual structure. In addition, the dilated convolution is also employed in the LD-Net to improve the accuracy of detection results. Experimental results based on benchmarks show that the proposed approach gets better performance in accuracy and memory occupied. The accuracy of the model on ICDAR dataset is 95.7% and the memory of the model occupies 39.7MB, which outperforms the existing methods.
Similar content being viewed by others
Explore related subjects
Discover the latest articles, news and stories from top researchers in related subjects.References
Bhowmik S, Kundu S, Sarkar R (2020) BINYAS: A complex document layout analysis system. Multimedia Tools Appl, pp 1–34
Binmakhashen GM, Mahmoud SA (2019) Document layout analysis: A comprehensive survey. ACM Comput Surv 52(6):1–36
Breuel T (2002) Two geometric algorithms for layout analysis. In: Proc ACM Int Workshop Doc Anal Syst, Princeton, USA, pp 188–199
Breuel T (2008) The OCRopus open source OCR system. In: Proc IS&T/SPIE 20th Annu Symp, San Jose, California, USA, pp 0F1–0F15
Bukhari SS, Shafait F, Breuel T (2011) Improved document image segmentation algorithm using multiresolution morphology. In: SPIE document recognition and retrieval XVIII, DRR’11, San Francisco, USA, pp 78740D–78740D
Bukhari S, Shafait F, Breuel T (2013) Towards generic text-line extraction. In: Proc Int Conf Document Anal Recognit (ICDAR), Washington, pp 748–752
Bukhari S, Shafait F, Breuel T (2013) Coupled snakelets for curled text-line segmentation from warped document images. Int J Doc Anal Recognit. (IJDAR) 16(1):33–53
Campos VB, Calvo-Zaragoza J, Toselli AH, Ruiz EV (2016) Sheet Music Statistical Layout Analysis. In: Proc 14th Int Conf Frontiers Handwriting Recognit (ICFHR), Shenzhen, China, pp 313–318
Chang F, Chu S-Y, Chen C-Y (2005) Chinese document layout analysis using adaptive regrouping strategy. Pattern Recognit 38:261–271
Dai-Ton H, Duc-Dung N, Duc-Hieu L (2016) An, adaptive over-split and merge algorithm for page segmentation. Pattern Recogn Lett 80:137–143
De R, Chakraborty A, Sarkar R (2020) Document image binarization using dual discriminator generative adversarial networks. IEEE Signal Process Lett 27:1090–1094
Gao L, Yi X, Jiang Z, Hao L, Tang Z (2017) ICDAR 2017 competition on page object detection. In: Proc 14th IAPR Int Conf Document Anal Recognit (ICDAR), Kyoto, Japan, pp 141–1422
Hesham AM, Rashwan MA, Al-Barhamtoshy HM, Abdou SM, Badr AA, Farag I (2017) Arabic document layout analysis. Pattern Anal Appl 20:1275–1287
Howard AG, Zhu M, Chen B, Kalenichenko D, Wang W, Weyand T, Andreetto M, Adam H (2017) Mobilenets: Efficient convolutional neural networks for mobile vision applications. arXiv:1704.04861
Kasar T, Barlas P, Adam S, Chatelain C, Paquet T (2013) Learning to detect tables in scanned document images using line information. In: Proc Int Conf Document Anal Recognit (ICDAR), pp 1185–1189
Koci E, Thiele M, Lehner W, Romero O (2018) Table recognition in spreadsheets via a graph representation. In: IAPR international workshop on document analysis systems (DAS). IEEE, Vienna, Austria, pp 139–144
Le VP, Nayef N, Visani M, Ogier J, Tran CD (2015) Text and non-text segmentation based on connected component features. In: Proc Int Conf Document Anal Recognit (ICDAR), Tunis, pp 1096–1100
Li Y, Zou Y, Ma J (2018) DeepLayout: A semantic segmentation approach to page layout analysis. In: Proc Int Conf Intell Comput, Bengaluru, India, pp 266–277
Min W, Fan M, Guo X, Han Q (2018) A new approach to track multiple vehicles with the combination of robust detection and two classifiers. IEEE Trans Intell Trans Syst 19:174–186
Moysset B, Messina R (2019) Are 2d-lstm really dead for offline text recognition. Int J Document Anal Recognit (IJDAR) 22:1–16
Nayef N, Ogier J (2015) Text zone classification using unsupervised feature learning. In: Proc Int Conf Document Anal Recognit (ICDAR), Tunis, pp 776–780
Nguyen NV, Rigaud C, Burie JC (2019) Comic MTL: optimized multi-task learning for comic book image analysis. Int J Document Anal Recognit (IJDAR) 22:265–284
Niu Y, Wen J, Zhong P, Xue Y (2019) A Hybrid, R-BILSTM-C neural network based text steganalysis. IEEE Signal Process Lett 26(12):1907–1911
Oliveira DAB, Viana PM (2017) Fast CNN-based document layout analysis. In: Proc IEEE Conf Comput Vis Pattern Recog, Waikiki, USA, pp 1173–1180
Otsu N (1979) Threshold selection method from gray-level histograms. IEEE Trans Syst Man Cybern SMC-9(1):62–66
Phillips I (1995) User’s reference manual, cd-rom, uw-iii document image database-iii
Qin X, Zhou Y, He Z, Wang Y, Tang Z (2017) A Faster R-CNN based method for comic characters face detection. In: Proc Int Conf Document Anal Recognit (ICDAR), Kyoto, Japan, pp 1074–1080
Royer E, Bouchara F (2017) Guiding text image keypoints extraction through layout analysis. In: Proc Int Conf Document Anal Recognit (ICDAR), Kyoto, Japan, pp 9–14
Selvaraju RR, Cogswell M, Das A, Vedantam R, Parikh D, Batra D (2017) Grad-CAM: Visual explanations from deep networks via gradient-based localization. In: Proc IEEE Int Conf Comput Vis, pp 618–626
Tran TA, Na IS, Kim SH (2016) Page segmentation using minimum homogeneity algorithm and adaptive mathematical morphology. Int J Doc Anal Recognit (IJDAR) 19(3):191–209
Tran TA, Na IS, Kim SH (2017) A robust system for document layout analysis using multilevel homogeneity structure. Expert Syst Appl 85:99–113
Tran DN, Tran TA, Oh A, Kim SH, Na IS (2005) Table detection from document image using vertical arrangement of text blocks. Int J Contents 11(4):77–85
Wang Q, Min W, He D, Zou S, Huang T, Zhang Y, Liu R (2020) Discriminative fine-grained network for vehicle re-identification using two-stage re-ranking. Sci China Inf Sci. https://doi.org/10.1007/385s11432-019-2811-8
Wong K, Casey R, Wahl F (1982) Document analysis systems. IBM J Res Dev 26(6):647–656
Yang J, Kim H, Kwak H, Kim I (2019) HanFont: large-scale adaptive Hangul font recognizer using CNN and font clustering. Int J Document Anal Recognit (IJDAR) 22:407–416
Yi X, Gao L, Liao Y, Zhang X, Liu R, Jiang Z (2017) CNN based page object detection in document images. In: Proc Int Conf Document Anal Recognit (ICDAR), Kyoto, Japan, pp 230–235
Yu F, Koltun V (2016) Multi-scale context aggregation by dilated convolutions. In: Proc Int Conf Learn Representations
Zhang X, Zhou X, Lin M, Sun J (2018) ShuffleNet: An extremely efficient convolutional neural network for mobile devices. In: Proc Conf Computer Vision and Pattern Recognition (CVPR), Salt Lake, pp 6848–6856
Acknowledgments
This work was supported by the National Natural Science Foundation of China (Grant No. 62076117 and No. 62166026), the Natural Science Foundation of Jiangxi Province, China (Grant No. 20161ACB20004) and Jiangxi Key Laboratory of Smart City (Grant No. 20192BCD40002).
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Zhao, H., Min, W., Wang, Q. et al. Memory-efficient document layout analysis method using LD-net. Multimed Tools Appl 82, 4371–4386 (2023). https://doi.org/10.1007/s11042-022-12497-9
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11042-022-12497-9