A computational approach for printed document forensics using SURF and ORB features

Kumar, Munish; Gupta, Surbhi; Mohan, Neeraj

doi:10.1007/s00500-020-04733-x

Cited by 18 publications

(16 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Inspired by this solution, several works have followed a similar path. Some examples are the Gray Level Co-occurrence Matrices (GLCMs) features from Mikkilineni et al [5]- [7], [12] and Ferreira et al [35], features from Distance Transform from Deng et al [9], features from Discrete Cosine Transform from Jiang et al [11], statistics of GLCMs, residual noise and Wavelet Transform features from Tsai and Liu [13], Tsai et al [15], [18] and Elkasrawi and Shafait [14], ad-hoc texture descriptors from Joshi and Khanna [20], [23], SURF and ORB features from Kumar et al [53], and geometric distortions signatures from Jain et al [54]. Finally, the use of deep neural networks from Ferreira et al [19] and their extension from Joshi et al [21] proved their ability to learn better the features from the data itself when sufficient data is used for training.…”

Section: Related Workmentioning

confidence: 99%

Ensembling Shallow Siamese Neural Network Architectures for Printed Documents Verification in Data-Scarcity Scenarios

Ferreira¹,

Purnekar

Barni

2021

IEEE Access

View full text Add to dashboard Cite

The popularity of printing devices has multiplied the diffusion of printed documents, raising concerns regarding the security and integrity of their content. The same device that prints reliable contracts, newspapers, and others, can also be used for malicious purposes, such as printing fake money, forging fake contracts, and produce illegal packaging, thus calling for the development of image forensics techniques to pinpoint criminal printed materials and trace back to their origin. In this work, we address the source linking problem of printed color documents by treating it as a verification problem. Specifically, we aim at deciding if two documents have been printed by the same printer or not. To achieve this goal, and to cope with the data scarcity deriving from the difficulty of gathering massive amounts of printed and scanned documents, we propose to use an ensemble of Siamese Neural Networks, with unique architectures expressly designed to work with a small training dataset. As a further unique feature, the proposed approach is suited to work in an open set scenario, where the printers used to produce the documents analyzed at the test time are not included in the training set. Results obtained under both open and closed set conditions, with a thorough comparison with available baseline methods, highlight the validity of the proposed solution and its capability to work in real-world settings.

show abstract

Section: Related Workmentioning

confidence: 99%

Ensembling Shallow Siamese Neural Network Architectures for Printed Documents Verification in Data-Scarcity Scenarios

Ferreira¹,

Purnekar

Barni

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…The scale invariance is realized by constructing Gaussian pyramid. 20 ORB adds an orientation component to FAST by utilizing an intensity centroid cloud mechanism. The centroid is found by moments of patch as in equation ( 1).…”

Section: Algorithm Descriptionmentioning

confidence: 99%

Autonomous recognition technology of carrier robot on various terrain environment

Dong

Song²,

Luan³

et al. 2021

Proceedings of the Institution of Mechanical Engineers, Part D:

View full text Add to dashboard Cite

Field carrier robot can help people carry a certain amount of necessary materials in mountaineering, exploration and other fields, and even help field soldiers carry a certain amount of armaments. However, due to the complexity of the field environment, it is difficult for a robot with a single locomotion mechanism to complete such work. A carrier robot with leg/caterpillar/wheel is proposed. The robot can recognize different terrain environment, so as to select the appropriate locomotion mechanism, and can avoid obstacles and plan the path at the same time. We optimize the traditional ORB algorithm to extract the features of the environment image. The random forest classifier is used to classify the terrain, so as to start the corresponding locomotion mechanism. In addition, the extracted ORB feature points are also used for stereo matching to achieve the depth recovery for environmental information. Based on the analysis of two path planning algorithms of DWA and artificial potential field, an artificial potential field algorithm based on DWA is proposed for robot local path planning. In order to verify the effectiveness of these methods, experiments are carried out on the switching for three locomotion mechanisms and path planning in different practical environments. The results show that when the terrain changes, the robot can effectively switch the locomotion mechanism, and can run stably in different environments.

show abstract

“…In the scope of the document understanding and recognition these algorithms are also playing an important role. They are used for document template matching [4], forensics checks [5] and even for the recognition of the characters [6].…”

Section: Introductionmentioning

confidence: 99%

Tiny CNN for feature point description for document analysis: approach and dataset

Sheshkus¹,

Chirvonaya²,

Arlazarov³

2022

Computer Optics

View full text Add to dashboard Cite

In this paper, we study the problem of feature points description in the context of document analysis and template matching. Our study shows that specific training data is required for the task especially if we are to train a lightweight neural network that will be usable on devices with limited computational resources. In this paper, we construct and provide a dataset of photo and synthetically generated images and a method of training patches generation from it. We prove the effectiveness of this data by training a lightweight neural network and show how it performs in both general and documents patches matching. The training was done on the provided dataset in comparison with HPatches training dataset and for the testing, we solve HPatches testing framework tasks and template matching task on two publicly available datasets with various documents pictured on complex backgrounds: MIDV-500 and MIDV-2019.

show abstract

A computational approach for printed document forensics using SURF and ORB features

Cited by 18 publications

References 22 publications

Ensembling Shallow Siamese Neural Network Architectures for Printed Documents Verification in Data-Scarcity Scenarios

Ensembling Shallow Siamese Neural Network Architectures for Printed Documents Verification in Data-Scarcity Scenarios

Autonomous recognition technology of carrier robot on various terrain environment

Tiny CNN for feature point description for document analysis: approach and dataset

Contact Info

Product

Resources

About