Showing papers by "Michael Blumenstein published in 2010"
09 Jun 2010
TL;DR: An overview and discussion of pen trajectory recovery methods developed to date is presented and the temporal order of the strokes or the pen trajectory is shown to be more promising for recovery.
Abstract: On-line handwriting recognition systems are usually better than their off-line counterparts thanks to the accessibility of dynamic information such as stroke order, velocity, acceleration, and pressure. Whilst the exact value of velocity as well as acceleration or pressure is unlikely to be recoverable, the temporal order of the strokes or the pen trajectory is shown to be more promising for recovery. The published experimental results suggest that the recovered pen trajectory information actually improves the off-line recognition accuracy. This paper presents an overview and discussion of pen trajectory recovery methods developed to date.
43 citations
16 Nov 2010
TL;DR: In this work, the performance of two feature extraction techniques, the Modified Direction Feature (MDF) and the gradient feature are compared on the basis of similar experimental settings and results indicated that an average error rate as low as 15.03% could be obtained using the gradient Feature and SVMs.
Abstract: Feature extraction is an important process in off-line signature verification. In this work, the performance of two feature extraction techniques, the Modified Direction Feature (MDF) and the gradient feature are compared on the basis of similar experimental settings. In addition, the performance of Support Vector Machines (SVMs) and the squared Mahalanobis distance classifier employing the Gradient Feature are also compared and reported. Without using forgeries for training, experimental results indicated that an average error rate as low as 15.03% could be obtained using the gradient feature and SVMs.
38 citations
16 Nov 2010
TL;DR: This paper focuses on the second scenario, which aims at performance evaluation of off-line signature verification systems on a newly-created large dataset that comprises genuine, simulated signatures produced by unskilled imitators or random signatures (genuine signatures from other writers).
Abstract: The objective of this competition (4NSigComp2010) is to ascertain the performance of automatic off-line signature verifiers to evaluate recent technology developments in the areas of document analysis and machine learning. The current paper focuses on the second scenario, which aims at performance evaluation of off-line signature verification systems on a newly-created large dataset that comprises genuine, simulated signatures produced by unskilled imitators or random signatures (genuine signatures from other writers). Ten systems were evaluated, and some interesting results are presented in terms of accuracy and execution time. The top ranking system attained an overall error of 8.94%. This result interestingly correlates with the top ranking accuracy achieved in a previous signature verification competition at ICDAR 2009.
32 citations
16 Nov 2010
TL;DR: This paper presents the 8 competition proposals for competitions to be held in the framework of the 12th International Conference on Frontiers in Handwriting Recognition (ICFHR2010) with lists of competition organizers and lists of participating systems and approaches.
Abstract: The great success and high number of participants in pattern recognition related competitions last years show an important improvement of recognition and classification approaches. This success is unconceivable without the availability of huge datasets of real world data. We have invited for proposals for competitions to be held in the framework of the 12th International Conference on Frontiers in Handwriting Recognition (ICFHR2010). These competitions should aim at evaluating the performance of algorithms and methods for a particular task of Handwriting Recognition. Eight different teams composed of more than one group have submitted their proposals. The subjects of these propositions cover the field of research of handwriting recognition from pre-processing over handwritten document analysis to handwriting text/word recognition. These competitions represent an overview of current research topics and frontiers in handwriting document analysis and recognition. Only 5 competitions have received enough participants (we have defined the threshold to 3 systems) to present their evaluation at the ICFHR 2010. This paper presents the 8 competition proposals with lists of competition organizers and lists of participating systems and approaches.
9 citations
25 Jul 2010
TL;DR: Following on site data collection and analysis it was found that the Wave Pack system provides a low cost, robust, reliable and accurate system for measuring continuous wave height and period from a low elevation video camera aimed at the target beach under a wide range of wave conditions.
Abstract: This paper presents a new system for measuring breakzone locations, breaking wave height and wave periods across the surfzone from a digital video sequence. The system (Wave Pack) aims to provide real-time measurement of breaking and re-breaking wave heights and wave periods using low mounted video camera installations. Following on site data collection and analysis it was found that the Wave Pack system provides a low cost, robust, reliable and accurate system for measuring continuous wave height and period from a low elevation video camera aimed at the target beach under a wide range of wave conditions. These tests have verified the accuracy of Wave Pack in comparison to existing systems.
6 citations
01 Jan 2010
TL;DR: In this paper, a case-based reasoning (CBR) method was used to predict long-term condition ratings of bridge elements based on the outcome of step 1 using Time Delay Neural Networks (TDNNs) and improving the prediction accuracy of step 2 by employing CBR.
Abstract: A bridge is principally designed to have a long service life. However, due to number factors, it could fail prematurely, and could cause loss of human life. In order to ensure the optimum bridge serviceability, systematic asset management is essential for effective decision-making of maintenance, repair and rehabilitation (MRR (2) predicting long-term condition ratings based on the outcome of Step 1 using Time Delay Neural Networks (TDNNs); and (3) improving long-term prediction accuracy of Step 2 by employing Case-based Reasoning (CBR). This paper mainly focuses on the first two steps of the research. Promising results are reported for the reliable long-term prediction of bridge element performance.
4 citations
01 Jan 2010
3 citations
01 Jan 2010
TL;DR: The CoastSAFE Alive project as mentioned in this paper is a real-time and forecast hazard and risk assessment on Australian and USA beaches using camera imagery collected from 140 Coastalwatch cameras in Australia, Hawaii, California and Texas.
Abstract: Rip currents are the major cause of beach rescues and drowning in beach-going countries In Australia they are responsible for 95% of surf rescues and most surf drownings In an effort to substantially improve our ability to detect, monitor and assess rips, as well as associated waves and other hazards CoastalCOMS is developing technologies that will provide real time and forecast hazard and risk assessment on Australian and USA beaches During the past year CoastalCOMS has collaborated with Coastalwatch, Surf Life Saving Australia, Surf Life Saving Queensland and Griffith University in the CoastSAFE Alive project that seeks to extend the utility of the camera imagery CoastalCOMS, the R&D spinout of Coastalwatch, utilises information from 140 Coastalwatch beach cameras in Australia, Hawaii, California and Texas Imagery from these cameras is analysed to provide information on breaker wave height and period, beach state, including presence and location of rips, shoreline position and number of people on the beach The CoastSAFE Alive project is integrating camera information with WW3, waverider and ANNA wave characteristics, weather condition, tide and lifeguard observation to provide both real time and forecast beach hazard and risk assessments In September 2009 the hazard and risk assessments were critically reviewed by beach managers and senior lifeguards at an International Workshop on the Gold Coast, Queensland This paper will present an overview of the CoastSAFE Alive project and application of the hazard and risk assessments to Australian and US beaches
2 citations
13 Dec 2010
TL;DR: A novel segmentation technique for segmenting neural cells in three dimensional images is presented and accuracy rates over 90% are reported on a data set of 100 images containing over 130 neural cells and subsequently validated using a novelData set of 64 neurons.
Abstract: Segmentation of neural cells in three dimensional fluorescence microscopy images is a challenging image processing problem. In addition to being important to neurobiologists, accurate segmentation is a vital component of an automated image processing system. Due to the complexity of the data, particularly the extreme irregularity in neural cell shape, generic segmentation techniques do not perform well. This paper presents a novel segmentation technique for segmenting neural cells in three dimensional images. Accuracy rates of over 90% are reported on a data set of 100 images containing over 130 neural cells and subsequently validated using a novel data set of 64 neurons.