Skip to main content

Ignazio Gallo

Università degli Studi dell'Insubria, Dipartimento di Informatica e Comunicazione, Faculty Member

Followers

112

Following

8

Co-authors

2

Public Views

I received my 'Laurea' (Master degree) in Computer Science from University of Milan, Italy, in 1998.

From 1998 to 2002, I worked at the National Research Council (CNR) of Milan for the Artificial Intelligence and Soft Computing Laboratory.

From 2002 to 2003, I worked at Department of Computer Science and Communication at the University of Insubria, Varese, in a project dealing with simulators in a distributed environment.

Since 2004, I am an Assistant Professor at the University of Insubria.

Research interests are in Image Processing, Pattern Recognition, Neural Computing and Computer Vision.

less

Zhengrong Liang

Zoubeida Messali

Mohd Amri Bin Md Yunus

Universiti Teknologi Malaysia - UTM

Francoise PEYRIN

INSA Lyon

The Australian National University

Manish Kumar Bajpai

"Tor Vergata" University of Rome

mahdi faramarzi

Universiti Teknologi Malaysia - UTM

InterestsView All (7)

Uploads

Papers by Ignazio Gallo

A Local and Iterative Neural Reconstruction Algorithm for Cone-Beam Data

Abstract. This work presents a new neural algorithm designed for the reconstruction of tomographi... more Abstract. This work presents a new neural algorithm designed for the reconstruction of tomographic images from Cone Beam data. The algorithm is iterative and based on a set of neural networks that are working locally and sequentially. The algorithm was compared with the iterative ART algorithm and the filtered backprojection (FBP) method. The results show how the proposed algorithm is much more accurate even in the presence of noise and under conditions of lack of data. 1

Cross-modal Speaker Verification and Recognition: A Multilingual Perspective

2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

Recent years have seen a surge in finding association between faces and voices within a cross-mod... more Recent years have seen a surge in finding association between faces and voices within a cross-modal biometric application along with speaker recognition. Inspired from this, we introduce a challenging task in establishing association between faces and voices across multiple languages spoken by the same set of persons. The aim of this paper is to answer two closely related questions: "Is face-voice association language independent?" and "Can a speaker be recognized irrespective of the spoken language?". These two questions are important to understand effectiveness and to boost development of multilingual biometric systems. To answer these, we collected a Multilingual AudioVisual dataset, containing human speech clips of 154 identities with 3 language annotations extracted from various videos uploaded online. Extensive experiments on the two splits of the proposed dataset have been performed to investigate and answer these novel research questions that clearly point out the relevance of the multilingual problem.

Digital privacy: Replacing pedestrians from Google Street View images

Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012), 2012

Given the lack of modern techniques to ensure the digital privacy of individuals, we want to pave... more Given the lack of modern techniques to ensure the digital privacy of individuals, we want to pave the way for a new approach to make pedestrians in cityscape images anonymous. To address these concerns, we propose an automated method to replace any unknown pedestrian with another one which is extracted from a controlled and authorized dataset. The techniques used up to now to make people anonymous are based mainly on the blurring of people's faces, but even so it is possible to trace the identity of the subject starting from his clothing, personal items, hairstyle, the place and time where the photo was taken. The proposed method aims to make the pedestrians completely anonymous, and consists of four phases: firstly we identify the area where the pedestrian is located, we separate the pedestrian from the background, we select the most similar pedestrian from a controlled dataset and subsequently we substitute it. Our case study is Google Street View because it is one of the onli...

E-Fairs: a Cyber-Physical System for Aggregation and Economy of Scale in e-Commerce

2018 IEEE 4th International Forum on Research and Technology for Society and Industry (RTSI)

In recent years, the e-commerce arena has deeply changed because of the advent of new business mo... more In recent years, the e-commerce arena has deeply changed because of the advent of new business models and the growing weight of huge global actors like Amazon. Some business models create competition between users, and the product price tends to rise (e.g., online auctions); other models, including group-buying, make users cooperate, and the price tends to go down. The present study extends the group-buying model and proposes a cyber-physical system called e-fair, in which both sellers and buyers are grouped to negotiate on a specific product or service. E-fairs minimize the global purchase price and the shipping resources respectively with the aggregation of demand and supply as well as origins and destinations. E-fairs aggregate sellers and buyers, sources and destinations in what we call double-side aggregation. As the aggregation regards independent actors, which do not trust each other and join e-fairs in dribs and drabs, we employ a promising distributed technology as the blockchain to make the aggregation. We validated the e-fair model through a system prototype and a simulator and understood how economies of scale apply to e-fairs in different usage scenarios.

Text categorization of commercial Web pages

In this paper we describe a new on-line document categorization strategy that can be integrated w... more In this paper we describe a new on-line document categorization strategy that can be integrated within Web applications. A salient aspect is the use of neural learning in both representation and classification tasks. Within text documents conceived as images, the regions of interest (RoI) containing information meaningful for categorization are identified with the support of a supervised neural network. Text within RoI is represented according to a simple solution that consider the first K words in the text and code them properly. A Kohonen Self-Organizing Map (SOM) is applied to cluster documents that are subsequently labelled by applying a simple majority voting mechanism. Solutions adopted were evaluated by conducting experiments within the context of on-line price comparison services. Results obtained demontrate that the overall classification strategy is able to categorize documents satisfectorily taking into account the high variability of Web pages.

Image and Text fusion for UPMC Food-101 using BERT and CNNs

2020 35th International Conference on Image and Vision Computing New Zealand (IVCNZ), 2020

The modern digital world is becoming more and more multimodal. Looking on the internet, images ar... more The modern digital world is becoming more and more multimodal. Looking on the internet, images are often associated with the text, so classification problems with these two modalities are very common. In this paper, we examine multimodal classification using textual information and visual representations of the same concept. We investigate two main basic methods to perform multimodal fusion and adapt them with stacking techniques to better handle this type of problem. Here, we use UPMC Food-101, which is a difficult and noisy multimodal dataset that well represents this category of multimodal problems. Our results show that the proposed early fusion technique combined with a stacking-based approach exceeds the state of the art on the dataset used.

Are These Birds Similar: Learning Branched Networks for Fine-grained Representations

2019 International Conference on Image and Vision Computing New Zealand (IVCNZ)

Fine-grained image classification is a challenging task due to the presence of hierarchical coars... more Fine-grained image classification is a challenging task due to the presence of hierarchical coarse-to-fine-grained distribution in the dataset. Generally, parts are used to discriminate various objects in fine-grained datasets, however, not all parts are beneficial and indispensable. In recent years, natural language descriptions are used to obtain information on discriminative parts of the object. This paper leverages on natural language description and proposes a strategy for learning the joint representation of natural language description and images using a two-branch network with multiple layers to improve the fine-grained classification task. Extensive experiments show that our approach gains significant improvements in accuracy for the fine-grained image classification task. Furthermore, our method achieves new state-of-the-art results on the CUB-200-2011 dataset.

Improving the Efficient Neural Architecture Search via Rewarding Modifications

2020 35th International Conference on Image and Vision Computing New Zealand (IVCNZ), 2020

Nowadays, a challenge for the scientific community concerning deep learning is to design architec... more Nowadays, a challenge for the scientific community concerning deep learning is to design architectural models to obtain the best performance on specific data sets. Building effective models is not a trivial task and it can be very time-consuming if done manually. Neural Architecture Search (NAS) has achieved remarkable results in deep learning applications in the past few years. It involves training a recurrent neural network (RNN) controller using Reinforcement Learning (RL) to automatically generate architectures. Efficient Neural Architecture Search (ENAS) was created to address the prohibitively expensive computational complexity of NAS using weight sharing. In this paper we propose Improved-ENAS (I-ENAS), a further improvement of ENAS that augments the reinforcement learning training method by modifying the reward of each tested architecture according to the results obtained in previously tested architectures. We have conducted many experiments on different public domain datase...

Key Sample Point Selection: An Improvement of Shape Context Algorithm in Image Retrieval

Signal Processing, Pattern Recognition and Applications, 2010

In this work we defined a new algorithm in the field of Content Based Image Retrieval. The Shape ... more In this work we defined a new algorithm in the field of Content Based Image Retrieval. The Shape Context Algorithm presents a promising solution to the Shape Analysis problem however its use is strongly limited by the high demand of time and space due to the elevated number of Sample Points required. The new algorithm proposed in this study aims to improve the original Shape Context algorithm's performance modifying some its relevant parts; furthermore, it was evaluated in term of accuracy, computational time and space. The salient aspects of our algorithm are: a new strategy for the Sample Points selection and a center of mass angle approximation technique in the phase of the shape description computation. We want to reduce the number of Sample Points required by the original algorithm in order to attempt to improve the efficiency in real applications.

Embedded Textual Content for Document Image Classification with Convolutional Neural Networks

Proceedings of the 2016 ACM Symposium on Document Engineering - DocEng '16, 2016

Integration of scale-space filtering and neural techniques for high resolution remote sensing image classification

Contextual classification methods, which require the extraction of complex spatial information ov... more

Robust Angle Invariant 1D Barcode Detection

2013 2nd Iapr Asian Conference on Pattern Recognition, Nov 8, 2013

Barcode reading mobile applications that identify products from pictures taken using mobile devic... more Barcode reading mobile applications that identify products from pictures taken using mobile devices are widely used by customers to perform online price comparisons or to access reviews written by others. Most of the currently available barcode reading approaches focus on decoding degraded barcodes and treat the underlying barcode detection task as a side problem that can be addressed using appropriate object detection methods. However, the majority of modern mobile devices do not meet the minimum working requirements of complex general purpose object detection algorithms and most of the efficient specifically designed barcode detection algorithms require user interaction to work properly. In this paper, we present a novel method for barcode detection in camera captured images based on a supervised machine learning algorithm that identifies onedimensional barcodes in the two-dimensional Hough Transform space. Our model is angle invariant, requires no user interaction and can be executed on a modern mobile device. It achieves excellent results for two standard one-dimensional barcode datasets: WWU Muenster Barcode Database and ArTe-Lab 1D Medium Barcode Dataset. Moreover, we prove that it is possible to enhance the overall performance of a state-of-the-art barcode reading algorithm by combining it with our detection method.

A Web-based solution to manage distributed discrete event simulations

SINPL (Simulator Integration Platform) is an open-source software platform supporting the integra... more SINPL (Simulator Integration Platform) is an open-source software platform supporting the integration of existing simulators in a distributed Web-based environment and the management of simulation experiments. In the present work we introduce the platform architecture and we discuss the role of the various modules composing the platform focusing on their visualization capabilities. Such capabilities aim at controlling and analyzing simulation experiments by means of the implemented 2D and 3D graphical user interfaces.

Soft Categorization and Annotation of Images with Radial Basis Function Networks

Visapp, 2009

This work focuses on fast approaches for image retrieval and classification by employing simple f... more This work focuses on fast approaches for image retrieval and classification by employing simple features to build image signatures. For this purpose a neural model for soft classification and automatic image annotation is proposed. The salient aspects of this solution are: a) the employment of a Radial Basis Function Network built on top of an image retrieval distance metric b) a soft learning strategy for annotation handling. Experiments have been conducted on a subset of the Corel image dataset for evaluation and comparative analysis.

Neural Image Restoration for Decoding 1-D Barcodes using Common Camera Phones

Visapp, 2010

Abstract The existing open-source libraries for 1-D barcodes recognition are not able to recogniz... more Abstract The existing open-source libraries for 1-D barcodes recognition are not able to recognize the codes from images acquired using simple devices without autofocus or macro function. In this article we present an improvement of an existing algorithm for recognizing 1-D barcodes using camera phones with and without autofocus. The multilayer feedforward neural network based on backpropagation algorithm is used for image restoration in order to improve the selected algorithm. Performances of the proposed algorithm were compared ...

A local and iterative neural reconstruction algorithm for cone-beam data

Proceedings of Spie the International Society For Optical Engineering, Mar 1, 2010

This work presents a new neural algorithm designed for the reconstruction of tomographic images f... more This work presents a new neural algorithm designed for the reconstruction of tomographic images from Cone Beam data. The algorithm is iterative and based on a set of neural networks that are working locally and sequentially. The algorithm was compared with the iterative ART algorithm and the filtered backprojection (FBP) method. The results show how the proposed algorithm is much more accurate even in the presence of noise and under conditions of lack of data.

Assigning Automatic Regularization Parameters in Image Restoration

Visapp, 2009

This work aims to define and experimentally evaluate an adaptive strategy based on neural learnin... more This work aims to define and experimentally evaluate an adaptive strategy based on neural learning to select an appropriate regularization parameter to restore a degraded image. It is well known that selecting an appropriate regularization parameter is very difficult in regularized method. To solve this problem, we propose a novel method to construct the regularization parameter function through a training concept using a supervised neural network in an attempt to overcome the limitations of trial and error and curve fitting procedures. The proposed solution is not included within a particular restoration algorithm. The results of our experiments indicate that this method may yield a model that can be generalised to restore never seen images.

Neural Adaptive Restoration of Computed Radiography Images

IFMBE Proceedings, 2009

Aim of this work is to experimentally investigate the potential of a novel technique for CR image... more Aim of this work is to experimentally investigate the potential of a novel technique for CR image restoration which make use of gradient descent algorithm to minimize a local error function derived from the conventional global constrained error measure adopted within regularization approaches. Results of preliminary experiments show that the proposed restoration algorithm is promising for medical imaging restoration and could be useful in limiting x-ray dose absorbed by patients.

A New Algorithm for Dense Two-Frame Stereo Correspondence

This work aims at defining a new method for matching correspondences in stereoscopic image analys... more This work aims at defining a new method for matching correspondences in stereoscopic image analysis. The salient aspects of the method are-an explicit representation of occlusions driving the overall matching process and the use of neural adaptive technique in disparity computation. In particular, based on the taxonomy proposed by Scharstein and Szelinsky, the dense stereo matching process has been divided into three tasks: matching cost computation, aggregation of local evidence and computation of disparity values. Within the second phase a new strategy has been introduced in an attempt to improve reliability in computing disparity. An experiment was conducted to evaluate the solutions proposed The experiment is based on an analysis of test images including data with a ground truth disparity map

Semi-blind image restoration using a local neural approach

Proceedings of the Fifth Iasted International Conference on Signal Processing Pattern Recognition and Applications, Feb 13, 2008

This work aims to define and experimentally evaluate an iterative strategy based on neural learni... more This work aims to define and experimentally evaluate an iterative strategy based on neural learning for semi-blind image restoration in the presence of blur and noise. Salient aspects of the proposed strategy are the use of a local error function derived from the conventional global constrained error measure and the assignment of a separate regularization parameter to each image pixel based on local gray level variance. This method can be viewed as a neural strategy where the pixels of the restored image are the synapse's weights the neural network tries to modify during learning to minimize the output error measurement. The method was experimentally evaluated in terms of restoration quality and speed using test images synthetically degraded and increasingly corrupted. To investigate whether the strategy can be considered an alternative to neural restoration procedures, the results were compared with those obtained by well known Hopfield-based restoration approaches. Results obtained show that our method performs significantly better and faster than other models considered.

A Local and Iterative Neural Reconstruction Algorithm for Cone-Beam Data

Abstract. This work presents a new neural algorithm designed for the reconstruction of tomographi... more Abstract. This work presents a new neural algorithm designed for the reconstruction of tomographic images from Cone Beam data. The algorithm is iterative and based on a set of neural networks that are working locally and sequentially. The algorithm was compared with the iterative ART algorithm and the filtered backprojection (FBP) method. The results show how the proposed algorithm is much more accurate even in the presence of noise and under conditions of lack of data. 1

Cross-modal Speaker Verification and Recognition: A Multilingual Perspective

2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

Recent years have seen a surge in finding association between faces and voices within a cross-mod... more Recent years have seen a surge in finding association between faces and voices within a cross-modal biometric application along with speaker recognition. Inspired from this, we introduce a challenging task in establishing association between faces and voices across multiple languages spoken by the same set of persons. The aim of this paper is to answer two closely related questions: "Is face-voice association language independent?" and "Can a speaker be recognized irrespective of the spoken language?". These two questions are important to understand effectiveness and to boost development of multilingual biometric systems. To answer these, we collected a Multilingual AudioVisual dataset, containing human speech clips of 154 identities with 3 language annotations extracted from various videos uploaded online. Extensive experiments on the two splits of the proposed dataset have been performed to investigate and answer these novel research questions that clearly point out the relevance of the multilingual problem.

Digital privacy: Replacing pedestrians from Google Street View images

Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012), 2012

Given the lack of modern techniques to ensure the digital privacy of individuals, we want to pave... more Given the lack of modern techniques to ensure the digital privacy of individuals, we want to pave the way for a new approach to make pedestrians in cityscape images anonymous. To address these concerns, we propose an automated method to replace any unknown pedestrian with another one which is extracted from a controlled and authorized dataset. The techniques used up to now to make people anonymous are based mainly on the blurring of people's faces, but even so it is possible to trace the identity of the subject starting from his clothing, personal items, hairstyle, the place and time where the photo was taken. The proposed method aims to make the pedestrians completely anonymous, and consists of four phases: firstly we identify the area where the pedestrian is located, we separate the pedestrian from the background, we select the most similar pedestrian from a controlled dataset and subsequently we substitute it. Our case study is Google Street View because it is one of the onli...

E-Fairs: a Cyber-Physical System for Aggregation and Economy of Scale in e-Commerce

2018 IEEE 4th International Forum on Research and Technology for Society and Industry (RTSI)

In recent years, the e-commerce arena has deeply changed because of the advent of new business mo... more In recent years, the e-commerce arena has deeply changed because of the advent of new business models and the growing weight of huge global actors like Amazon. Some business models create competition between users, and the product price tends to rise (e.g., online auctions); other models, including group-buying, make users cooperate, and the price tends to go down. The present study extends the group-buying model and proposes a cyber-physical system called e-fair, in which both sellers and buyers are grouped to negotiate on a specific product or service. E-fairs minimize the global purchase price and the shipping resources respectively with the aggregation of demand and supply as well as origins and destinations. E-fairs aggregate sellers and buyers, sources and destinations in what we call double-side aggregation. As the aggregation regards independent actors, which do not trust each other and join e-fairs in dribs and drabs, we employ a promising distributed technology as the blockchain to make the aggregation. We validated the e-fair model through a system prototype and a simulator and understood how economies of scale apply to e-fairs in different usage scenarios.

Text categorization of commercial Web pages

In this paper we describe a new on-line document categorization strategy that can be integrated w... more In this paper we describe a new on-line document categorization strategy that can be integrated within Web applications. A salient aspect is the use of neural learning in both representation and classification tasks. Within text documents conceived as images, the regions of interest (RoI) containing information meaningful for categorization are identified with the support of a supervised neural network. Text within RoI is represented according to a simple solution that consider the first K words in the text and code them properly. A Kohonen Self-Organizing Map (SOM) is applied to cluster documents that are subsequently labelled by applying a simple majority voting mechanism. Solutions adopted were evaluated by conducting experiments within the context of on-line price comparison services. Results obtained demontrate that the overall classification strategy is able to categorize documents satisfectorily taking into account the high variability of Web pages.

Image and Text fusion for UPMC Food-101 using BERT and CNNs

2020 35th International Conference on Image and Vision Computing New Zealand (IVCNZ), 2020

The modern digital world is becoming more and more multimodal. Looking on the internet, images ar... more The modern digital world is becoming more and more multimodal. Looking on the internet, images are often associated with the text, so classification problems with these two modalities are very common. In this paper, we examine multimodal classification using textual information and visual representations of the same concept. We investigate two main basic methods to perform multimodal fusion and adapt them with stacking techniques to better handle this type of problem. Here, we use UPMC Food-101, which is a difficult and noisy multimodal dataset that well represents this category of multimodal problems. Our results show that the proposed early fusion technique combined with a stacking-based approach exceeds the state of the art on the dataset used.

Are These Birds Similar: Learning Branched Networks for Fine-grained Representations

2019 International Conference on Image and Vision Computing New Zealand (IVCNZ)

Fine-grained image classification is a challenging task due to the presence of hierarchical coars... more Fine-grained image classification is a challenging task due to the presence of hierarchical coarse-to-fine-grained distribution in the dataset. Generally, parts are used to discriminate various objects in fine-grained datasets, however, not all parts are beneficial and indispensable. In recent years, natural language descriptions are used to obtain information on discriminative parts of the object. This paper leverages on natural language description and proposes a strategy for learning the joint representation of natural language description and images using a two-branch network with multiple layers to improve the fine-grained classification task. Extensive experiments show that our approach gains significant improvements in accuracy for the fine-grained image classification task. Furthermore, our method achieves new state-of-the-art results on the CUB-200-2011 dataset.

Improving the Efficient Neural Architecture Search via Rewarding Modifications

2020 35th International Conference on Image and Vision Computing New Zealand (IVCNZ), 2020

Nowadays, a challenge for the scientific community concerning deep learning is to design architec... more Nowadays, a challenge for the scientific community concerning deep learning is to design architectural models to obtain the best performance on specific data sets. Building effective models is not a trivial task and it can be very time-consuming if done manually. Neural Architecture Search (NAS) has achieved remarkable results in deep learning applications in the past few years. It involves training a recurrent neural network (RNN) controller using Reinforcement Learning (RL) to automatically generate architectures. Efficient Neural Architecture Search (ENAS) was created to address the prohibitively expensive computational complexity of NAS using weight sharing. In this paper we propose Improved-ENAS (I-ENAS), a further improvement of ENAS that augments the reinforcement learning training method by modifying the reward of each tested architecture according to the results obtained in previously tested architectures. We have conducted many experiments on different public domain datase...

Key Sample Point Selection: An Improvement of Shape Context Algorithm in Image Retrieval

Signal Processing, Pattern Recognition and Applications, 2010

In this work we defined a new algorithm in the field of Content Based Image Retrieval. The Shape ... more In this work we defined a new algorithm in the field of Content Based Image Retrieval. The Shape Context Algorithm presents a promising solution to the Shape Analysis problem however its use is strongly limited by the high demand of time and space due to the elevated number of Sample Points required. The new algorithm proposed in this study aims to improve the original Shape Context algorithm's performance modifying some its relevant parts; furthermore, it was evaluated in term of accuracy, computational time and space. The salient aspects of our algorithm are: a new strategy for the Sample Points selection and a center of mass angle approximation technique in the phase of the shape description computation. We want to reduce the number of Sample Points required by the original algorithm in order to attempt to improve the efficiency in real applications.

Embedded Textual Content for Document Image Classification with Convolutional Neural Networks

Proceedings of the 2016 ACM Symposium on Document Engineering - DocEng '16, 2016

Integration of scale-space filtering and neural techniques for high resolution remote sensing image classification

Contextual classification methods, which require the extraction of complex spatial information ov... more

Robust Angle Invariant 1D Barcode Detection

2013 2nd Iapr Asian Conference on Pattern Recognition, Nov 8, 2013

Barcode reading mobile applications that identify products from pictures taken using mobile devic... more Barcode reading mobile applications that identify products from pictures taken using mobile devices are widely used by customers to perform online price comparisons or to access reviews written by others. Most of the currently available barcode reading approaches focus on decoding degraded barcodes and treat the underlying barcode detection task as a side problem that can be addressed using appropriate object detection methods. However, the majority of modern mobile devices do not meet the minimum working requirements of complex general purpose object detection algorithms and most of the efficient specifically designed barcode detection algorithms require user interaction to work properly. In this paper, we present a novel method for barcode detection in camera captured images based on a supervised machine learning algorithm that identifies onedimensional barcodes in the two-dimensional Hough Transform space. Our model is angle invariant, requires no user interaction and can be executed on a modern mobile device. It achieves excellent results for two standard one-dimensional barcode datasets: WWU Muenster Barcode Database and ArTe-Lab 1D Medium Barcode Dataset. Moreover, we prove that it is possible to enhance the overall performance of a state-of-the-art barcode reading algorithm by combining it with our detection method.

A Web-based solution to manage distributed discrete event simulations

SINPL (Simulator Integration Platform) is an open-source software platform supporting the integra... more SINPL (Simulator Integration Platform) is an open-source software platform supporting the integration of existing simulators in a distributed Web-based environment and the management of simulation experiments. In the present work we introduce the platform architecture and we discuss the role of the various modules composing the platform focusing on their visualization capabilities. Such capabilities aim at controlling and analyzing simulation experiments by means of the implemented 2D and 3D graphical user interfaces.

Soft Categorization and Annotation of Images with Radial Basis Function Networks

Visapp, 2009

This work focuses on fast approaches for image retrieval and classification by employing simple f... more This work focuses on fast approaches for image retrieval and classification by employing simple features to build image signatures. For this purpose a neural model for soft classification and automatic image annotation is proposed. The salient aspects of this solution are: a) the employment of a Radial Basis Function Network built on top of an image retrieval distance metric b) a soft learning strategy for annotation handling. Experiments have been conducted on a subset of the Corel image dataset for evaluation and comparative analysis.

Neural Image Restoration for Decoding 1-D Barcodes using Common Camera Phones

Visapp, 2010

Abstract The existing open-source libraries for 1-D barcodes recognition are not able to recogniz... more Abstract The existing open-source libraries for 1-D barcodes recognition are not able to recognize the codes from images acquired using simple devices without autofocus or macro function. In this article we present an improvement of an existing algorithm for recognizing 1-D barcodes using camera phones with and without autofocus. The multilayer feedforward neural network based on backpropagation algorithm is used for image restoration in order to improve the selected algorithm. Performances of the proposed algorithm were compared ...

A local and iterative neural reconstruction algorithm for cone-beam data

Proceedings of Spie the International Society For Optical Engineering, Mar 1, 2010

This work presents a new neural algorithm designed for the reconstruction of tomographic images f... more This work presents a new neural algorithm designed for the reconstruction of tomographic images from Cone Beam data. The algorithm is iterative and based on a set of neural networks that are working locally and sequentially. The algorithm was compared with the iterative ART algorithm and the filtered backprojection (FBP) method. The results show how the proposed algorithm is much more accurate even in the presence of noise and under conditions of lack of data.

Assigning Automatic Regularization Parameters in Image Restoration

Visapp, 2009

This work aims to define and experimentally evaluate an adaptive strategy based on neural learnin... more This work aims to define and experimentally evaluate an adaptive strategy based on neural learning to select an appropriate regularization parameter to restore a degraded image. It is well known that selecting an appropriate regularization parameter is very difficult in regularized method. To solve this problem, we propose a novel method to construct the regularization parameter function through a training concept using a supervised neural network in an attempt to overcome the limitations of trial and error and curve fitting procedures. The proposed solution is not included within a particular restoration algorithm. The results of our experiments indicate that this method may yield a model that can be generalised to restore never seen images.

Neural Adaptive Restoration of Computed Radiography Images

IFMBE Proceedings, 2009

Aim of this work is to experimentally investigate the potential of a novel technique for CR image... more Aim of this work is to experimentally investigate the potential of a novel technique for CR image restoration which make use of gradient descent algorithm to minimize a local error function derived from the conventional global constrained error measure adopted within regularization approaches. Results of preliminary experiments show that the proposed restoration algorithm is promising for medical imaging restoration and could be useful in limiting x-ray dose absorbed by patients.

A New Algorithm for Dense Two-Frame Stereo Correspondence

This work aims at defining a new method for matching correspondences in stereoscopic image analys... more This work aims at defining a new method for matching correspondences in stereoscopic image analysis. The salient aspects of the method are-an explicit representation of occlusions driving the overall matching process and the use of neural adaptive technique in disparity computation. In particular, based on the taxonomy proposed by Scharstein and Szelinsky, the dense stereo matching process has been divided into three tasks: matching cost computation, aggregation of local evidence and computation of disparity values. Within the second phase a new strategy has been introduced in an attempt to improve reliability in computing disparity. An experiment was conducted to evaluate the solutions proposed The experiment is based on an analysis of test images including data with a ground truth disparity map

Semi-blind image restoration using a local neural approach

Proceedings of the Fifth Iasted International Conference on Signal Processing Pattern Recognition and Applications, Feb 13, 2008

This work aims to define and experimentally evaluate an iterative strategy based on neural learni... more This work aims to define and experimentally evaluate an iterative strategy based on neural learning for semi-blind image restoration in the presence of blur and noise. Salient aspects of the proposed strategy are the use of a local error function derived from the conventional global constrained error measure and the assignment of a separate regularization parameter to each image pixel based on local gray level variance. This method can be viewed as a neural strategy where the pixels of the restored image are the synapse's weights the neural network tries to modify during learning to minimize the output error measurement. The method was experimentally evaluated in terms of restoration quality and speed using test images synthetically degraded and increasingly corrupted. To investigate whether the strategy can be considered an alternative to neural restoration procedures, the results were compared with those obtained by well known Hopfield-based restoration approaches. Results obtained show that our method performs significantly better and faster than other models considered.