Search results for "Pattern recognition"

Large scale semi-supervised image segmentation with active queries

2011

A semiautomatic procedure to generate classification maps of remote sensing images is proposed. Starting from a hierarchical unsupervised classification, the algorithm exploits the few available labeled pixels to assign each cluster to the most probable class. For a given amount of labeled pixels, the algorithm returns a classified segmentation map, along with confidence levels of class membership for each pixel. Active learning methods are used to select the most informative samples to increase confidence in the class membership. Experiments on a AVIRIS hyperspectral image confirm the effectiveness of the method, especially when used with active learning query functions and spatial regular…

Classification based on Iterative Object Symmetry Transform

2004

The paper shows an application of a new operator named the iterated object transform (IOT) for cell classification. The IOT has the ability to grasp the internal structure of a digital object and this feature can be usefully applied to discriminate structured images. This is the case of cells representing chondrocytes in bone tissue, giarda protozoan, and myeloid leukaemia. A tree classifier allows us to discriminate the three classes with a good accuracy.

Multi-Temporal Image Classification with Kernels

2009

An improved MSD-based method for PD defects classification

2006

The new proposed method of pattern recognition is based on the application of Multi-resolution Signal Decomposition (MSD) technique of wavelet transform. This technique has showed off interesting properties in capturing the embedded horizontal, vertical and diagonal variations within an image obtained from the PD pattern in a separable form. This feature was exploited to identify in the PD pattern's MSD, relative at various family of partial discharge sources, some detail images typical of a single discharge phenomenon. The classification of a generic PD phenomenon is feasible through a comparison between its detail images and the detail images typical of a single discharge phenomenon. Test…

A note on the iterative object symmetry transform

2004

This paper introduces a new operator named the iterated object transform that is computed by combining the object symmetry transform with the morphological operator erosion. This new operator has been applied on both binary and gray levels images showing the ability to grasp the internal structure of a digital object. We present also some experiments on artificial and real images and potential applications.

A Support Vector Domain Description Approach to Supervised Classification of Remote Sensing Images

2007

This paper addresses the problem of supervised classification of remote sensing images in the presence of incomplete (nonexhaustive) training sets. The problem is analyzed according to two different perspectives: 1) description and recognition of a specific land-cover class by using single-class classifiers and 2) solution of multiclass problems with single-class classification techniques. In this framework, we analyze different one-class classifiers and introduce in the remote sensing community the support vector domain description method (SVDD). The SVDD is a kernel-based method that exhibits intrinsic regularization ability and robustness versus low numbers of high-dimensional samples. T…

Graph matching for efficient classifiers adaptation

2011

In this work we present an adaptation algorithm focused on the description of the measurement changes under different acquisition conditions. The adaptation is carried out by transforming the manifold in the first observation conditions into the corresponding manifold in the second. The eventually non-linear transform is based on vector quantization and graph matching. The transfer learning mapping is defined in an unsupervised manner. Once this mapping has been defined, the labeled samples in the first are projected into the second domain, thus allowing the application of any classifier in the transformed domain. Experiments on VHR series of images show the validity of the proposed method …

Encoding Invariances in Remote Sensing Image Classification With SVM

2013

This letter introduces a simple method for including invariances in support-vector-machine (SVM) remote sensing image classification. We design explicit invariant SVMs to deal with the particular characteristics of remote sensing images. The problem of including data invariances can be viewed as a problem of encoding prior knowledge, which translates into incorporating informative support vectors (SVs) that better describe the classification problem. The proposed method essentially generates new (synthetic) SVs from the obtained by training a standard SVM with the available labeled samples. Then, original and transformed SVs are used for training the virtual SVM introduced in this letter. W…

Incorporating in vivo and ex vivo NMR sources of information for modeling robust brain tumor classifiers

2010

The purpose of this paper is to investigate the potential and limitations of using multimodal sources of information coming from in vivo NMR and ex vivo NMR data for detecting brain tumors. Supervised pattern recognition methods, whose performance directly depends on the prior available observations used in building them, are proposed. We show that high resolution magic angle spinning (HR-MAS) data act as complementary information for classifying magnetic resonance spectroscopic imaging (MRSI) data. In particularly, when considering rare brain tumors, since it is unlikely to acquire sufficient cases to define their metabolite profiles using only in vivo NMR information, HR-MAS can support t…

A non-supervised approach to locate and to measure the nuchal translucency by means of wavelet analysis and neural networks

2017

Ultrasound imaging is a well known noninvasive way to evaluate various diseases during the prenatal age. In particular, the thickness measure of the nuchal transucency is strictly correlated with pathologies like trisomy 13, 18 and 21. For a correct investigation, the methodology needs mid-sagittal sections and the proposed approach is based on wavelet analysis and neural network classifiers to locate components useful to identify mid-sagittal planes. To evaluate the performance and the robustness of the methodology, real clinical ultrasound images were considered, obtaining an average error of at most 0.3 millimeters in 97.4% of the cases.