Malignant melanoma is one of the most fatal forms of skin cancer. It has also become increasingly common, especially among white-skinned people exposed to the sun. Early detection of melanoma is essential to raise survival rates, since its detection at an early stage can be helpful and curable. Working out the dermoscopic clinical features (pigment network and lesion borders) of melanoma is a vital step for dermatologists, who require an accurate method of reaching the correct clinical diagnosis, and ensure the right area receives the correct treatment. These structures are considered one of the main keys that refer to melanoma or non-melanoma disease. However, determining these clinical features can be a time-consuming, subjective (even for trained clinicians) and challenging task for several reasons: lesions vary considerably in size and colour, low contrast between an affected area and the surrounding healthy skin, especially in early stages, and the presence of several elements such as hair, reflections, oils and air bubbles on almost all images. This thesis aims to provide an accurate, robust and reliable automated dermoscopy image analysis technique, to facilitate the early detection of malignant melanoma disease. In particular, four innovative methods are proposed for region segmentation and classification, including two for pigmented region segmentation, one for pigment network detection, and one for lesion classification. In terms of boundary delineation, four pre-processing operations, including Gabor filter, image sharpening, Sobel filter and image inpainting methods are integrated in the segmentation approach to delete unwanted objects (noise), and enhance the appearance of the lesion boundaries in the image. The lesion border segmentation is performed using two alternative approaches. The Fuzzy C-means and the Markov Random Field approaches detect the lesion boundary by repeating the labeling of pixels in all clusters, as a first method. Whereas, the Particle Swarm Optimization with the Markov Random Field method achieves greater accuracy for the same aim by combining them in the second method to perform a local search and reassign all image pixels to its cluster properly. With respect to the pigment network detection, the aforementioned pre-processing method is applied, in order to remove most of the hair while keeping the image information and increase the visibility of the pigment network structures. Therefore, a Gabor filter with connected component analysis are used to detect the pigment network lines, before several features are extracted and fed to the Artificial Neural Network as a classifier algorithm. In the lesion classification approach, the K-means is applied to the segmented lesion to separate it into homogeneous clusters, where important features are extracted; then, an Artificial Neural Network with Radial Basis Functions is trained by representative features to classify the given lesion as melanoma or not. The strong experimental results of the lesion border segmentation methods including Fuzzy C-means with Markov Random Field and the combination between the Particle Swarm Optimization and Markov Random Field, achieved an average accuracy of 94.00% , 94.74% respectively. Whereas, the lesion classification stage by using extracted features form pigment network structures and segmented lesions achieved an average accuracy of 90.1% , 95.97% respectively. The results for the entire experiment were obtained using a public database PH2 comprising 200 images. The results were then compared with existing methods in the literature, which have demonstrated that our proposed approach is accurate, robust, and efficient in the segmentation of the lesion boundary, in addition to its classification.

The goal of this master thesis was to propose a suitable strategy to detect and classify objects of interest in mammogram images. A part of this goal was to implement an experimentation framework, that will be used for data preparation, model training and comparison. Patch and full-image versions of the dataset were used in the analysis. Initialisation with weights that were pretrained on the images from other domain improved classifier performance. ResNet-34 had better AUC scores on the test set that ResNet-18. Semi-supervised training using entropy minimisation has no significant improvement over the supervised one. The thesis includes the visualisation of the network predictions and the analysis of the knowledge representation of the classier. The achieved results for a patch version of the dataset are comparable to the results of another article that utilised the same test set. For a full-image dataset the results of the training were suboptimal. 1

Corneal diseases are one of the major causes of visual impairment and blindness worldwide. Used for diagnoses, a laser confocal microscope provides a sequence of images, at incremental depths, of the various corneal layers and structures. From these, ophthalmologists can extract clinical information on the state of health of a patient’s cornea. However, many factors impede ophthalmologists in forming diagnoses starting with the large number and variable quality of the individual images (blurring, non-uniform illumination within images, variable illumination between images and noise), and there are also difficulties posed for automatic processing caused by eye movements in both lateral and axial directions during the scanning process. Aiding ophthalmologists working with long sequences of corneal image requires the development of new algorithms which enhance, correctly order and register the corneal images within a sequence. The novel algorithms devised for this purpose and presented in this thesis are divided into four main categories. The first is enhancement to reduce the problems within individual images. The second is automatic image classification to identify which part of the cornea each image belongs to, when they may not be in the correct sequence. The third is automatic reordering of the images to place the images in the right sequence. The fourth is automatic registration of the images with each other. A flexible application called CORNEASYS has been developed and implemented using MATLAB and the C language to provide and run all the algorithms and methods presented in this thesis. CORNEASYS offers users a collection of all the proposed approaches and algorithms in this thesis in one platform package. CORNEASYS also provides a facility to help the research team and Ophthalmologists, who are in discussions to determine future system requirements which meet clinicians’ needs. / The data and image files accompanying this thesis are not available online.

Face perception is impaired in a variety of ways in older adults, but the mechanisms underlying these changes remain unclear. A central theme of this dissertation is that task performance is constrained by two factors intrinsic to the observer: sources of random variability -- internal noise -- and the efficiency with which task-relevant stimulus information is utilized. This thesis uses several behavioural, psychophysical methods to examine how age-related changes in one or both of these factors affect face processing. Chapter 2 used the classification image (CI) method to characterize the spatial sampling patterns of younger and older observers performing a face discrimination task. Compared to younger adults, older adults used information in the eye/brow region less consistently and instead relied on relatively less informative regions such as the forehead. The differences in CIs accounted for the lower absolute efficiency that was found in older observers. Chapter 3 estimated internal noise and calculation efficiency by measuring threshold-vs.-noise (TvN) curves and response consistency in a face discrimination task. Compared to younger observers, older observers had higher additive internal noise and lower calculation efficiency, but the magnitude of multiplicative internal noise did not differ between age groups. Previous studies have shown that younger adults have a bias to rely on horizontal structure to discriminate and identify faces, and the magnitude of this so-called horizontal bias is correlated with identification accuracy. The experiments in Chapter 4 measured horizontal bias in younger and older adults, and found that age differences in horizontal bias account for some, but not all, of the age difference in face identification accuracy. In summary, my work demonstrates that additive (but not multiplicative) internal noise is greater in older adults, and that they are less efficient at sampling information that is conveyed by structure at different locations and orientations in a face. / Dissertation / Doctor of Science (PhD) / Our experience of the visual environment results from perceptual processes in the brain. Many of these processes change with age, such as our ability to identify someone from a photograph of their face. Performance is influenced by both random variability, or "noise", within the observer and how efficiently we use task-relevant information in the visual environment. By systematically manipulating the amount of available stimulus information I assessed the contribution of these factors to older adults' judgements of facial identity, and characterized the information on which these decisions are based. These experiments are the first to consider how face perception in older adults is constrained by the combined effects of internal noise and the efficiency with which the visual system utilizes various sources of information. The results provide a number of directions for future research in the fields of face perception and age-related changes in complex pattern vision.

Corneal diseases are one of the major causes of visual impairment and blindness worldwide. Used for diagnoses, a laser confocal microscope provides a sequence of images, at incremental depths, of the various corneal layers and structures. From these, ophthalmologists can extract clinical information on the state of health of a patient’s cornea. However, many factors impede ophthalmologists in forming diagnoses starting with the large number and variable quality of the individual images (blurring, non-uniform illumination within images, variable illumination between images and noise), and there are also difficulties posed for automatic processing caused by eye movements in both lateral and axial directions during the scanning process. Aiding ophthalmologists working with long sequences of corneal image requires the development of new algorithms which enhance, correctly order and register the corneal images within a sequence. The novel algorithms devised for this purpose and presented in this thesis are divided into four main categories. The first is enhancement to reduce the problems within individual images. The second is automatic image classification to identify which part of the cornea each image belongs to, when they may not be in the correct sequence. The third is automatic reordering of the images to place the images in the right sequence. The fourth is automatic registration of the images with each other. A flexible application called CORNEASYS has been developed and implemented using MATLAB and the C language to provide and run all the algorithms and methods presented in this thesis. CORNEASYS offers users a collection of all the proposed approaches and algorithms in this thesis in one platform package. CORNEASYS also provides a facility to help the research team and Ophthalmologists, who are in discussions to determine future system requirements which meet clinicians’ needs.

A good image representation is vital for image comparision and classification; it may affect the classification accuracy and efficiency. The purpose of this thesis was to explore novel and appropriate image representations. Another aim was to investigate these representations for image classification. Finally, novel features were examined for improving image classification accuracy. Images of interest to this thesis were textile design images. The motivation of analysing textile design images is to help designers browse images, fuel their creativity, and improve their design efficiency. In recent years, bag-of-words model has been shown to be a good base for image representation, and there have been many attempts to go beyond this representation. Bag-of-words models have been used frequently in the classification of image data, due to good performance and simplicity. “Words” in images can have different definitions and are obtained through steps of feature detection, feature description, and codeword calculation. The model represents an image as an orderless collection of local features. However, discarding the spatial relationships of local features limits the power of this model. This thesis exploited novel image representations, bag of shapes and region label graphs models, which were based on bag-of-words model. In both models, an image was represented by a collection of segmented regions, and each region was described by shape descriptors. In the latter model, graphs were constructed to capture the spatial information between groups of segmented regions and graph features were calculated based on some graph theory. Novel elements include use of MRFs to extract printed designs and woven patterns from textile images, utilisation of the extractions to form bag of shapes models, and construction of region label graphs to capture the spatial information. The extraction of textile designs was formulated as a pixel labelling problem. Algorithms for MRF optimisation and re-estimation were described and evaluated. A method for quantitative evaluation was presented and used to compare the performance of MRFs optimised using alpha-expansion and iterated conditional modes (ICM), both with and without parameter re-estimation. The results were used in the formation of the bag of shapes and region label graphs models. Bag of shapes model was a collection of MRFs' segmented regions, and the shape of each region was described with generic Fourier descriptors. Each image was represented as a bag of shapes. A simple yet competitive classification scheme based on nearest neighbour class-based matching was used. Classification performance was compared to that obtained when using bags of SIFT features. To capture the spatial information, region label graphs were constructed to obtain graph features. Regions with the same label were treated as a group and each group was associated uniquely with a vertex in an undirected, weighted graph. Each region group was represented as a bag of shape descriptors. Edges in the graph denoted either the extent to which the groups' regions were spatially adjacent or the dissimilarity of their respective bags of shapes. Series of unweighted graphs were obtained by removing edges in order of weight. Finally, an image was represented using its shape descriptors along with features derived from the chromatic numbers or domination numbers of the unweighted graphs and their complements. Linear SVM classifiers were used for classification. Experiments were implemented on data from Liberty Art Fabrics, which consisted of more than 10,000 complicated images mainly of printed textile designs and woven patterns. Experimental data was classified into seven classes manually by assigning each image a text descriptor based on content or design type. The seven classes were floral, paisley, stripe, leaf, geometric, spot, and check. The result showed that reasonable and interesting regions were obtained from MRF segmentation in which alpha-expansion with parameter re-estimation performs better than alpha-expansion without parameter re-estimation or ICM. This result was not only promising for textile CAD (Computer-Aided Design) to redesign the textile image, but also for image representation. It was also found that bag of shapes model based on MRF segmentation can obtain comparable classification accuracy with bag of SIFT features in the framework of nearest neighbour class-based matching. Finally, the result indicated that incorporation of graph features extracted by constructing region label graphs can improve the classification accuracy compared to both bag of shapes model and bag of SIFT models.

Thèse de doctorat : Mathématiques : Nice : 2004. / Thèse préparée à l'Inria Sophia Antipolis, projet Ariana. Bibliogr. p. 261-269.

This thesis deals with convolutional neural networks. It is a kind of deep neural networks that are presently widely used mainly for image recognition and natural language processing. The thesis describes specifics of convolutional neural networks in comparison with traditional neural networks and is focused on inner computations in the process of learning. Convolutional neural networks typically consist of a different types of layers of neurons and the core part of this thesis is to demonstrate computations of individual types of layers. Learning demonstrating program of a simple convolutional network was designed and implemented using own implementation of neural network. Validity of the implementation was tested by training models for solving a classification task. Experiments with different types of architectures were conducted and their performance was compared.

The PhD thesis deals with modern methods of image processing, especially image segmentation, classification and evaluation of parameters. It is focused primarily on processing medical images of soft tissues obtained by magnetic resonance tomography (MR) and microscopic images of tissues. It is easy to describe edges of the sought objects using of segmented images. The edges found can be useful for further processing of monitored object such as calculating the perimeter, surface and volume evaluation or even three-dimensional shape reconstruction. The proposed solutions can be used for the classification of healthy/unhealthy tissues in MR or other imaging. Application examples of the proposed segmentation methods are shown in this thesis. Research in the area of image segmentation is focused on methods based on solving partial differential equations. This is a modern method for image processing, often called the active contour method. It is of great advantage in the segmentation of real images degraded by noise with fuzzy edges and transitions between objects. The results of the thesis are methods proposed for automatic image segmentation and classification.

With the availability of inexpensive hardware and software, digital imaging has become an important medium of communication in our daily lives. A huge amount of digital images are being collected and become available through the internet and stored in various fields such as personal image collections, medical imaging, digital arts etc. Therefore, it is important to make sure that images are stored, searched and accessed in an efficient manner. The use of bag of visual words (BOW) model for modelling images based on local invariant features computed at interest point locations has become a standard choice for many computer vision tasks. Based on this promising model, this thesis investigates three main problems: natural scene classification, annotation and retrieval. Given an image, the task is to design a system that can determine to which class that image belongs to (classification), what semantic concepts it contain (annotation) and what images are most similar to (retrieval). This thesis contributes to scene classification by proposing a weighting approach, named keypoints density-based weighting method (KDW), to control the fusion of colour information and bag of visual words on spatial pyramid layout in a unified framework. Different configurations of BOW, integrated visual vocabularies and multiple image descriptors are investigated and analyzed. The proposed approaches are extensively evaluated over three well-known scene classification datasets with 6, 8 and 15 scene categories using 10-fold cross validation. The second contribution in this thesis, the scene annotation task, is to explore whether the integrated visual vocabularies generated for scene classification can be used to model the local semantic information of natural scenes. In this direction, image annotation is considered as a classification problem where images are partitioned into 10x10 fixed grid and each block, represented by BOW and different image descriptors, is classified into one of predefined semantic classes. An image is then represented by counting the percentage of every semantic concept detected in the image. Experimental results on 6 scene categories demonstrate the effectiveness of the proposed approach. Finally, this thesis further explores, with an extensive experimental work, the use of different configurations of the BOW for natural scene retrieval.