No abstract

No abstract

2D color barcodes have been introduced to obtain larger storage capabilities than traditional black and white barcodes. Unfortunately, the data density of color barcodes is substantially limited by the redundancy needed for correcting errors, which are due not only to geometric but also to chromatic distortions introduced by the printing and scanning process. The higher the expected error rate, the more redundancy is needed for avoiding failures in barcode reading, and thus, the lower the actual data density. Our work addresses this trade-off between reliability and data density in 2D color barcodes and aims at identifying the most effective algorithms, in terms of byte error rate and computational overhead, for decoding 2D color barcodes. In particular, we perform a thorough experimental study to identify the most suitable color classifiers for converting analog barcode cells to digital bit streams. To accomplish this task, we implemented a prototype capable of decoding 2D color barcodes by using different methods, including clustering algorithms and machine learning classifiers. We show that, even if state-of-art methods for color classification could be successfully used for decoding color barcodes in the desktop scenario, there is an emerging need for new color classification methods in the mobile scenario. In desktop scenarios, our experimental findings show that complex techniques, such as support vector machines, does not seem to pay off, as they do not achieve better accuracy in classifying color barcode cells. The lowest error rates are indeed obtained by means of clustering algorithms and probabilistic classifiers. From the computational viewpoint, classification with clustering seems to be the method of choice. In mobile scenarios, simple and efficient methods (in terms of computational time) such as the Euclidean and the K-means classifiers are not effective (in terms of error rate), while, more complex methods are effective but not efficient. Even if a few color barcode designs have been proposed in recent studies, to the best of our knowledge, there is no previous research that addresses a comparative and experimental analysis of clustering and machine learning methods for color classification in 2D color barcodes.

Abstract. The wide availability of on-board cameras in mobile devices and the increasing demand for higher capacity have recently sparked many new color barcode designs. Unfortunately, color barcodes are much more prone to errors than black and white barcodes, due to the chromatic distortions introduced in the printing and scanning process. This is a severe limitation: the higher the expected error rate, the more redundancy is needed for error correction (in order to avoid failures in barcode reading), and thus the lower the actual capacity achieved. Motivated by this, we design, engineer and experiment algorithms for decoding color barcodes with high accuracy. Besides tackling the general trade-off between error correction and data density, we address challenges that are specific to mobile scenarios and that make the problem much more complicated in practice. In particular, correcting chromatic distortions for barcode pictures taken from phone cameras appears to be a great challenge, since pictures taken from phone cameras present a very large variation in light conditions. We propose a new barcode decoding algorithm based on graph drawing methods, which is able to run in few seconds even on low-end computer architectures and to achieve nonetheless high accuracy in the recognition phase. The main idea of our algorithm is to perform color classification using force-directed graph drawing methods: barcode elements which are very close in color will attract each other, while elements that are very far will repulse each other.

No abstract

Handwritten Signature Verification (HSV) systems have been introduced to automatically verify the authenticity of a user signature. In offline systems, the handwritten signature (represented as an image) is taken from a scanned document, while in online systems, pen tablets are used to register signature dynamics (e.g., its position, pressure and velocity). In online HSV systems, signatures (including the signature dynamics) may be embedded into digital documents. Unfortunately, during their lifetime documents may be repeatedly printed and scanned (or faxed), and digital to paper conversions may result in loosing the signature dynamics. The main contribution of this work is a new HSV system for document signing and authentication. First, we illustrate how to verify handwritten signatures so that signature dynamics can be processed during verification of every type of document (both paper and digital documents). Secondly, we show how to embed features extracted from handwritten signatures within the documents themselves (by means of 2D barcodes), so that no remote signature database is needed. Thirdly, we propose a method for the verification of signature dynamics which is compatible to a wide range of mobile devices (in terms of computational overhead and verification accuracy) so that no special hardware is needed. We address the trade-off between discrimination capabilities of the system and the storage size of the signature model. Towards this end, we report the results of an experimental evaluation of our system on different handwritten signature datasets.