André R. S. Marçal

Comparison of Segmentation Methods for Melanoma Diagnosis in Dermoscopy Images

In this paper, we propose and evaluate six methods for the segmentation of skin lesions in dermoscopic images. This set includes some state of the art techniques which have been successfully used in many medical imaging problems (gradient vector flow (GVF) and the level set method of Chan et al.[(C-LS)]. It also includes a set of methods developed by the authors which were tailored to this particular application (adaptive thresholding (AT), adaptive snake (AS), EM level set (EM-LS), and fuzzy-based split-and-merge algorithm (FBSM)]. The segmentation methods were applied to 100 dermoscopic images and evaluated with four different metrics, using the segmentation result obtained by an experienced dermatologist as the ground truth. The best results were obtained by the AS and EM-LS methods, which are semi-supervised methods. The best fully automatic method was FBSM, with results only slightly worse than AS and EM-LS.

PH 2 - A dermoscopic image database for research and benchmarking

The increasing incidence of melanoma has recently promoted the development of computer-aided diagnosis systems for the classification of dermoscopic images. Unfortunately, the performance of such systems cannot be compared since they are evaluated in different sets of images by their authors and there are no public databases available to perform a fair evaluation of multiple systems. In this paper, a dermoscopic image database, called PH2, is presented. The PH2 database includes the manual segmentation, the clinical diagnosis, and the identification of several dermoscopic structures, performed by expert dermatologists, in a set of 200 dermoscopic images. The PH2 database will be made freely available for research and benchmarking purposes.

Optical music recognition: state-of-the-art and open issues

For centuries, music has been shared and remembered by two traditions: aural transmission and in the form of written documents normally called musical scores. Many of these scores exist in the form of unpublished manuscripts and hence they are in danger of being lost through the normal ravages of time. To preserve the music some form of typesetting or, ideally, a computer system that can automatically decode the symbolic images and create new scores is required. Programs analogous to optical character recognition systems called optical music recognition (OMR) systems have been under intensive development for many years. However, the results to date are far from ideal. Each of the proposed methods emphasizes different properties and therefore makes it difficult to effectively evaluate its competitive advantages. This article provides an overview of the literature concerning the automatic analysis of images of printed and handwritten musical scores. For self-containment and for the benefit of the reader, an introduction to OMR processing systems precedes the literature overview. The following study presents a reference scheme for any researcher wanting to compare new OMR algorithms against well-known ones.

Land cover update by supervised classification of segmented ASTER images

The revision of the 1995 land cover dataset for the Vale do Sousa region, in the northwest of Portugal, was carried out by supervised classification of a multi‐spectral image from the Advanced Spaceborne Thermal Emission and Reflectance Radiometer (ASTER) sensor. The nine reflective bands of ASTER were used, covering the spectral range from 0.52–2.43 µm. The image was initially ortho‐rectified and segmented into 51 186 objects, with an average object size of 135 pixels (about 3 ha). A total of 582 of these objects were identified for training nine land cover classes. The image was classified using an algorithm based on a fuzzy classifier, Support Vector Machines (SVM), K Nearest Neighbours (K‐NN) and a Logistic Discrimination (LD) classifier. The results from the classification were evaluated using a set of 277 validation sites, independently gathered. The overall accuracy was 44.6% for the fuzzy classifier, 70.5% for the SVM, 60.9% for the K‐NN and 72.2% for the LD classifier. The difficulty in discriminating between some of the forest land cover classes was examined by separability analysis and unsupervised classification with hierarchical clustering. The forest classes were found to overlap in the multi‐spectral space defined by the nine ASTER bands used.

Bayesian Hyperspectral Image Segmentation With Discriminative Class Learning

This paper introduces a new supervised technique to segment hyperspectral images: the Bayesian segmentation based on discriminative classification and on multilevel logistic (MLL) spatial prior. The approach is Bayesian and exploits both spectral and spatial information. Given a spectral vector, the posterior class probability distribution is modeled using multinomial logistic regression (MLR) which, being a discriminative model, allows to learn directly the boundaries between the decision regions and, thus, to successfully deal with high-dimensionality data. To control the machine complexity and, thus, its generalization capacity, the prior on the multinomial logistic vector is assumed to follow a componentwise independent Laplacian density. The vector of weights is computed via the fast sparse multinomial logistic regression (FSMLR), a variation of the sparse multinomial logistic regression (SMLR), conceived to deal with large data sets beyond the reach of the SMLR. To avoid the high computational complexity involved in estimating the Laplacian regularization parameter, we have also considered the Jeffreys prior, as it does not depend on any hyperparameter. The prior probability distribution on the class-label image is an MLL Markov-Gibbs distribution, which promotes segmentation results with equal neighboring class labels. The -expansion optimization algorithm, a powerful graph-cut-based integer optimization tool, is used to compute the maximum a posteriori segmentation. The effectiveness of the proposed methodology is illustrated by comparing its performance with the state-of-the-art methods on synthetic and real hyperspectral image data sets. The reported results give clear evidence of the relevance of using both spatial and spectral information in hyperspectral image segmentation.

Hierarchical clustering of multispectral images using combined spectral and spatial criteria

An agglomerative hierarchical clustering method, which uses both spectral and spatial information for the aggregation decision, is proposed here. The method is suitable for large multispectral images, provided that an unsupervised classification is previously applied. The method is tested on a synthetic image and on a satellite image of the coastal zone.

The use of texture for image classification of black & white air photographs

The use of black & white (B&W) air photographs for the production of historic land cover maps can be done by image classification, using additional texture features. In this paper we evaluate the importance of a number of parameters in the image classification process based on texture, such as the window size, angle and distance used to produce the texture features, the number of features used, the image quantization level and its spatial resolution. The evaluation was performed using five photographs from the 1950s. The influence of the classification method, the number of classes searched for in the images and the post‐processing tasks were also investigated. The effect of each of these parameters for the classification accuracy was evaluated by cross‐validation. The selection of the best parameters was performed based on the validation results, and also on the computation load involved for each case and the end user requirements. The final classification results were good (average accuracy of 85.7%, k = 0.809) and the method has proven to be useful for the production of historic land cover maps from B&W air photographs.

Comparison of Segmentation Methods for Automatic Diagnosis of Dermoscopy Images

Dermoscopy is a non-invasive diagnostic technique for the in vivo observation of pigmented skin lesions used in dermatology. There is currently a great interest in the prospects of automatic image analysis methods for dermoscopy, both to provide quantitative information about a lesion, which can be of relevance for the clinician, and as a stand alone early warning tool. The effective implementation of such a tool could lead to a reduction in the number of cases selected for exeresis, with obvious benefits both to the patients and to the health care system. The standard approach in automatic dermoscopic image analysis has usually three stages: (i) image segmentation, (ii) feature extraction and feature selection, (iii) lesion classification. This paper presents a comparison of segmentation methods applied to 50 dermoscopic image analysis, along with a clinical evaluation of each segmentation result performed by an experienced dermatologist.

Evaluation of Features for Leaf Discrimination

A number of shape features for automatic plant recognition based on digital image processing have been proposed by Pauwels et al. in 2009. A database with 15 classes and 171 leaf samples was considered for the evaluation of these measures using linear discriminant analysis and hierarchical clustering. The results obtained match the human visual shape perception with an overall accuracy of 87%.

Image Processing of Artificial Targets for Automatic Evaluation of Spray Quality

A fully automatic methodology based on image processing is proposed to evaluate the quality of spray application sampled by water-sensitive papers (WSP). The methods proposed permit a computation of the fraction of spray coverage, an evaluation of the homogeneity of the spray spatial spread at various scales and directions, and extraction of stain and droplet size range and distribution. This allows the number of droplets per unit area and the standard droplet size spectra factors to be computed. The methods were tested with a number of test samples scanned at different resolutions, proving to be effective in situations where there is high spray coverage in the WSP, thus with considerable overlap between stains. The most suitable scanning resolution was found to be 600 dpi. The results obtained by the image processing methods were successfully compared with a manual (visual) counting of stains in a test sample.