Figure 6 - uploaded by Zhongyuan Guo
Content may be subject to copyright.
Source publication
With the rapid development of information technology and the widespread use of the Internet, QR codes are widely used in all walks of life and have a profound impact on people's work and life. However, the QR code itself is likely to be printed and forged, which will cause serious economic losses and criminal offenses. Therefore, it is of great sig...
Context in source publication
Context 1
... a confusion matrix is used to measure the accuracy of classification model, the confusion matrix of LeNet, AlexNet, Min-Jen Tsai, and PSINet is shown in Figure 6, where the columns represent predicted labels and the rows represent true labels, the values on the diagonal indicates the number of correct prediction, and non-diagonal elements are the part of prediction errors. The higher the value on the diagonal of the confusion matrix, the more accurate the prediction result. ...
Similar publications
QR codes (short for Quick Response codes) were originally developed for use in the automotive industry to track factory inventories and logistics, but their popularity has expanded significantly in the past few years due to the widespread applications of smartphones and mobile phone cameras. QR codes can be used for a variety of purposes, including...
Citations
... However, its recognition performance will be significantly degenerated under the condition of mobile device acquisition. Guo (Guo et al., 2020) proposed a convolutional neural network for printer source identification of QR codes scanned by scanners, named PSINet, which is mainly composed of residual modules and achieved high identification accuracy on eight printers. This is because the scanner is specialised equipment, the QR code image scanned by scanner is clear and of high quality, so it is easy to extract the features that represent the attributes of the printers. ...
... In the field of printer source identification, the commonly used CNNs include AlexNet (Krizhevsky et al., 2012) and GoogLeNet (Szegedy et al., 2015) and ResNet18 (He et al., 2016). In addition, there are CNNs proposed based on the characteristics of printer source identification, including PDSI proposed by Tsai (Tsai et al., 2019) and PSINet proposed by Guo (Guo et al., 2020): PDSI was designed in the order of 5 × 5 conv1 layer，max pooling layer，5 × 5 conv2 layer, maxing pooling layer，5 × 5 conv3 layer, average pooling layer, fully connected layer and was mainly used for printer source identification of text documents collected by scanner and microscope, PSINet was designed in the order of 5 × 5 convolutional layer, average pooling layer, BRB1, average pooling layer, BRB2, average pooling layer, BRB3, global average pooling layer, and a fully connected layer, the kernel sizes of all average pooling layers are 5 × 5, and PSINet was mainly used to identify the printer source of QR codes collected by scanner. Hence, this paper makes comparison with the above deep learning methods to verify the effectiveness and superiority of the proposed method, and the number of layers and parameters of each CNN are shown in Table 10. ...
... (3) Since there are so many brands, models and numbers of printers in the real world, and new printers are emerging all the time, we will investigate effective solutions to quickly identify new printers, such as meta-learning based methods. (Guo et al., 2020) 58m55s 0.49 PDSI (Tsai et al., 2019) 17m12s 0.45 GoogLeNet (Szegedy et al., 2015) 5h48 m 2.31 ResNet18 (He et al., 2016) 7h8m 1.75 SE-BRB-Net 2h41 m 0.52 ...
Quick response codes are widely utilized in commodity anti-counterfeiting, and their traceability via smart-phones is an effective authentication method. However, quick response codes are susceptible to illegal duplication , and the images captured by smartphones are easily affected by changes in light and environmental noise, leading to unsatisfactory verification results. To address these issues, this paper introduces a novel approach by combining the squeeze-excitation attention module with bottleneck residual block. It presents a squeeze-excitation bottleneck residual network for printer source identification of quick response codes. The squeeze-excitation attention module pays more attention to the features that represent the printer attributes, reduces the interference of useless information, and has low computational consumption; while the bottleneck residual block has the advantages of few parameters, strong feature extraction capability, and good expandability. Thus, the performance can be improved effectively only with a small increasing in parameters. The experimental results verify that the proposed method achieves an accuracy of 98.77% under smartphone capture conditions, it outperforms other convolutional neural network-based methods in terms of identification accuracy. The deep learning model proposed in this paper can be generalized and applied to the printer source identification of paper content in the civil, criminal investigation and judicial fields.
... where each input image pixel f (x, y) is converted to the output image pixel g(x, y) based on the threshold value T [20]. ...
The advancement of Internet of Things (IoT) has enhanced the extensive usage of QR code images in various computer vision applications. Nonetheless, this has also brought forth several technical challenges. In particular, the logistics sorting system often encounters issues such as a low recognition rate and slow processing speed when dealing with QR code images under complex lighting conditions like uneven illumination. To address these difficulties, a method that focuses on achieving a fast adaptive binarization of QR code images through dynamic illumination equalization was proposed. First, an algorithm based on edge enhancement to obtain the position detection patterns within QR code images was applied, which enabled the acquisition of structural features in uneven illumination. Subsequently, QR code images with complex lighting conditions can achieve a fast adaptive binarization through dynamic illumination equalization. As for method validation, the experiments were performed on the two datasets that include QR code images influenced by strong light, weak light, and different shadow degrees. The results disclosed the benefits of the proposed method compared to the previous approaches; it produced superior recognition rates of 78.26–98.75% in various cases through commonly used decoders (Wechat and Zxing), with a faster processing speed of 0.0164 s/image, making it a proper method to satisfy real-time requirements in practical applications, such as a logistics sorting system.
... To ensure that the results are based solely on intrinsic signatures, extrinsic signature information such as watermarks is excluded from the datasets. The approach leverages a deep learning framework based on CNNs for image classification, and the data preprocessing follows a method similar to that used in supervised learning [21,22]. To avoid bias, related data are printed in nearly the same dimensions within the same scanner. ...
... Zhongyuan Guo [22] proposed using a customized bottleneck residual model to identify the source printer using black and white QR codes as datasets scanned at 300 dpi. Otsu's segmentation was used to convert the QR codes from grayscale to black and white to minimize variance, resulting in the whole dataset being converted into binary images. ...
QR codes (short for Quick Response codes) were originally developed for use in the automotive industry to track factory inventories and logistics, but their popularity has expanded significantly in the past few years due to the widespread applications of smartphones and mobile phone cameras. QR codes can be used for a variety of purposes, including tracking inventory, advertising, electronic ticketing, and mobile payments. Although they are convenient and widely used to store and share information, their accessibility also means they might be forged easily. Digital forensics can be used to recognize direct links of printed documents, including QR codes, which is important for the investigation of forged documents and the prosecution of forgers. The process involves using optical mechanisms to identify the relationship between source printers and the duplicates. Techniques regarding computer vision and machine learning, such as convolutional neural networks (CNNs), can be implemented to study and summarize statistical features in order to improve identification accuracy. This study implemented AlexNet, DenseNet201, GoogleNet, MobileNetv2, ResNet, VGG16, and other Pretrained CNN models for evaluating their abilities to predict the source printer of QR codes with a high level of accuracy. Among them, the customized CNN model demonstrated better results in identifying printed sources of grayscale and color QR codes with less computational power and training time.
... Population, in general, is becoming more conscious about the various methods and techniques used to solve various types of crimes including frauds and forgeries [5]. General population currently have a better understanding of how handwriting is different in each and every individual and can be easily recognized and matched to the original writer, which is especially important in the context of QDE [6]. ...
Printed documents are very frequently encountered in forensic cases as disputed or questioned documents. With increasing number of such cases, the printer inspection in has become a major requirement in questioned document examination in recent years considering the extensive use printers in document creation in comparison to handwritten papers. Also, the counterfeiting of printed documents by various printers have been recorded at a large scale in last two decades. In such cases, it is very much required by the investigators to examine and identify the type of printer used and to establish a linkage of the questioned document with the alleged printer. The current study is focused on the examination, analysis and identification of various types of printed documents on the basis of their specific characteristics including resolution, edge contrast, letter roughness and feathering formed from diverse printers to differentiate and classify them for forensic questioned document examination and to assist the forensic expert and questioned document examiner thorough such investigations. In this study 64 types of printers i.e. 32 inkjet printers and 32 printers comprise 1024 printed documents were collected for the analysis. The samples were analyzed using Olympus stereo microscopes SZX16 (4x magnification). Each printer has its uniqueness in printing documents, and this study aims to figure out that particular uniqueness at the documents after printing from the alleged printer. The observations revealed substantial differences in the letter’s characteristics. This study may be a useful to discriminate the documents printed from different types of printers.
... Bank checks, Anonymous letters, disputed contracts, wills, passports, petitions, threatening letters, and suicide notes are some examples [4]. In the last few decades, printing has become increasingly popular; variety of easily available and inexpensive printers are encountered at every work place and household now a days [1,5]. As per a study released by the Statistic Brain Research Centre in September 2016, "106 billions of printers are sold globally each year and as per data available the annual sales of inkjet printers and laser printers are18 billion USD and 30 billion USD, respectively" [6]. ...
Printed documents are very frequently encountered in forensic cases as disputed or questioned documents. With increasing number of such cases, the printer inspection in has become a major requirement in questioned document examination in recent years considering the extensive use printers in document creation in comparison to handwritten papers. Also, the counterfeiting of printed documents by various printers have been recorded at a large scale in last two decades. In such cases, it is very much required by the investigators to examine and identify the type of printer used and to establish a linkage of the questioned document with the alleged printer. The current study is focused on the examination, analysis and identification of various types of printed documents on the basis of their specific characteristics including resolution, edge contrast, letter roughness and feathering formed from diverse printers to differentiate and classify them for forensic questioned document examination and to assist the forensic expert and questioned document examiner thorough such investigations. In this study three types of printers i.e. two inkjet printers, two laser printers, and two different dot-matrix printers were used to collected the printed documents. The samples were analyzed using Projectina Inspec-8 and Stereomicroscope. Each printer has its uniqueness in printing documents, and this study aims to figure out that particular uniqueness at the documents after printing from the alleged printer. The observations revealed substantial differences in the letter’s characteristics. This study may be a useful to discriminate the documents printed from different types of printers.
Image forensics is an investigation of digital images to identify manipulations that have been done on them. Nowadays, due to the availability of different low-cost devices for capturing images, digital images are gaining quite a bit of popularity. It occurs frequently that these images are manipulated by mistake or on purpose, resulting in inaccurate information being presented by the image. There is a need to develop techniques to identify forgeries present in digital images used by the media, in court trials, and for maintaining visual records, since digital images are commonly used as evidence through the media, in court, and for maintaining record keeping. A detailed review of various image forgery detection techniques is presented in this article including comparisons between the various methods, pros and cons, and results obtained during the experimentation.
