This paper focuses on the problem of feature extraction and the classification of microvascular morphological types to aid esophageal cancer detection. We present a patch-based system with a hybrid SVM model with data augmentation for intraepithelial papillary capillary loop recognition. A greedy patch-generating algorithm and a specialized CNN named NBI-Net are designed to extract hierarchical...

We introduce an algorithm based on the morphological shared-weight neural network. Being nonlinear and translation-invariant, the MSNN can be used to create better generalization during face recognition. Feature extraction is performed on grayscale images using hit-miss transforms that are independent of gray-level shifts. The output is then learned by interacting with the classification proces...

This paper presents a new approach for large vocabulary Arabic speech recognition based on exploiting the morphological structures of the Arabic language. In this model, word discrimination is achieved by a hybrid analysis scheme, where vowels are described in detail while consonants are classifi ed according to broad phonetic classes. Different phonetic classifi cation strategies are used to d...

An iris recognition algorithm is implemented using morphological image processing, Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). The suggested occlusion removal technique is implemented using erosion and dilation operators, and compared against a training set without occlusion removal and one that only keeps one half of the iris. The resulting comparison shows that ...

Many accounts of the processing of morphologically complex words have been elaborated within the interactive activation model of word identification. Conceptually, this model adopts a "segmentation-through-recognition" approach to morphological decomposition, which assumes that a complex word activates representations of constituent morphemes as well as the representation of the whole word. How...

MORPHOLOGICAL BOUNDARY BASED SHAPE REPRESENTATION SCHEMES ON MOMENT INVARIANTS FOR CLASSIFICATION OF TEXTURES V. Venkata Krishna1 and M. Rama Bai2 1C.I.E.T, JNTUK, Kakinada, Rajahmundry, Andhra Pradesh, India, E-mail: [email protected] 2Dept. of CSE, M.G.I.T, JNTUH, Hyderabad, Andhra Pradesh, India, E-mail: [email protected] Efficient shape representation and recognition is an imp...

Major advances have been made in the past 10 yr in both the understanding of the biologic characteristics of acute nonlymphocytic leukemia and in the treatment of patients with this disease. Advances in the biologic characteristics include: a better understanding of the nature of leukemic cell proliferation and differentiation; a clearer description of the morphological, histochemical, and ultr...

This paper addresses the problem of speaker-independent connected numeral recognition over telephone lines. Increasing the vocabulary from digits (0-9) to numerals (0-99) opens for more user-friendly services, but it also introduces many new, language-specific problems. This paper investigates morphological, phonemic and allophonic variations in the pronunciation of numerals in Norwegian. If im...

In the analysis of natural language text a key step is named entity recognition, finding all complex noun phrases that denote persons, organizations, locations, and other entities designated by a name. In this paper we introduce the hunner open source language-independent named entity recognition system, and present results for Hungarian. When the input to hunner is already morphologically anal...

Connecting automatic speech recognizers (ASRs) and language analyzers is difficult since they may be based on differences in part-of-speech (POS) systems; the latter cannot directly analyze the outputs of the former. In addition, in unsegmented languages such as Japanese, the ASR outputs are likely to have different word segmentation from that of the language analyzer inputs because they are in...