Loretta Ichim

Unmanned Aerial Vehicle Systems for Remote Estimation of Flooded Areas Based on Complex Image Processing

Floods are natural disasters which cause the most economic damage at the global level. Therefore, flood monitoring and damage estimation are very important for the population, authorities and insurance companies. The paper proposes an original solution, based on a hybrid network and complex image processing, to this problem. As first novelty, a multilevel system, with two components, terrestrial and aerial, was proposed and designed by the authors as support for image acquisition from a delimited region. The terrestrial component contains a Ground Control Station, as a coordinator at distance, which communicates via the internet with more Ground Data Terminals, as a fixed nodes network for data acquisition and communication. The aerial component contains mobile nodes—fixed wing type UAVs. In order to evaluate flood damage, two tasks must be accomplished by the network: area coverage and image processing. The second novelty of the paper consists of texture analysis in a deep neural network, taking into account new criteria for feature selection and patch classification. Color and spatial information extracted from chromatic co-occurrence matrix and mass fractal dimension were used as well. Finally, the experimental results in a real mission demonstrate the validity of the proposed methodologies and the performances of the algorithms.

Characterization of Tumor Angiogenesis Using Fractal Measures

Tumors vascular networks are different from normal vascular networks, but the mechanisms underlying these differences are not known. Underlying these mechanisms may be key to improving the efficacy of the treatment of tumor. We studied possibility to apply two types of fractal measure: fractal dimension and succolarity for characterizing medical images. We find that fractal dimension value to the normal vasculature is smaller than the results for the tumor vasculature. Moreover, for applying succolarity the results do not vary considerable with the direction in normal vasculature, while for tumor vasculature the curves differs significantly. In conclusion, the results obtained show that the fractal measure is an important tool for analyzing medical images.

Two Dimensional Modeling and Fractal Characterization of Tumor Vascular Network

Tumor networks display percolation like scaling, representing the first evidence for a biological growth process whose key determinants are local substrate properties. In this paper we present a full characterization of a recently proposed model which reproduces the main features of the biological system, focusing on its dynamical properties, on the fractal properties of patterns, and on the percolative phase transition. We propose a simple model which reproduces many features of the biological system.

Improving Texture Based Classification of Aerial Images by Fractal Features

In this paper we propose an effective method of aerial image classification, which combines three types of features: color-based, statistical and fractal information. Two distinct phases were necessary for the CBIR system, which includes the classification algorithm: the learning phase and the classification phase. In the learning phase 5 different and efficient features were selected: entropy, contrast, homogeneity, mass fractal dimension and lacunarity. Also, three categories (classes) in CBIR were considered. The method of comparison, based on sub-images, improves the texture-based classification. A set of 100 aerial images from UAV was tested for establishing the rate of classification. The rate of 96% accurate classification, obtained as result, confirms the efficiency of the proposed method.

Intelligent Image Processing System for Detection and Segmentation of Regions of Interest in Retinal Images

The automatic detection, segmentation, localization, and evaluation of the optic disc, macula, exudates, and hemorrhages are very important for diagnosing retinal diseases. One of the difficulties in detecting such regions of interest (RoIs) with computer vision is their symmetries, e.g., between the optic disc and exudates and also between exudates and hemorrhages. This paper proposes an original, intelligent, and high-performing image processing system for the simultaneous detection and segmentation of retinal RoIs. The basic principles of the method are image decomposition in small boxes and local texture analysis. The processing flow contains three phases: preprocessing, learning, and operating. As a first novelty, we propose proper feature selection based on statistical analysis in confusion matrices for different feature types (extracted from a co-occurrence matrix, fractal type, and local binary patterns). Mainly, the selected features are chosen to differentiate between similar RoIs. The second novelty consists of local classifier fusion. To this end, the local classifiers associated with features are grouped in global classifiers corresponding to the RoIs. The local classifiers are based on minimum distances to the representatives of classes and the global classifiers are based on confidence intervals, weights, and a voting scheme. A deep convolutional neural network, based on supervised learning, for blood vessel segmentation is proposed in order to improve the RoI detection performance. Finally, the experimental results on real images from different databases demonstrate the rightness of our methodologies and algorithms.

Blood vessel segmentation in eye fundus images

This paper proposes a convolutional neural network architecture for blood vessel segmentation in retinal images. The network structure is designed on 7 layers using MatConvNet (three convolutional layers, two pooling layers, one dropout layer and a Softmax layer). The input data, selected from the DRIVE database, of the neural network is preprocessed in Matlab on Green channel. The retinal image was partitioned in patches of dimension 27 × 27 pixels using a sliding box algorithm. The whole network was trained using a number of 400,000 patches. The CNN was implemented using GPU Programming in MATLAB. The results are promising taking into account speed of processing and the simplicity of the network.

Using fractal dimension as discriminator of infected HeLa cells from spectrophotometric images

The paper presents an original experimental optical method to characterise, using fractal dimension, cell nuclei size distribution for virus-infected and non infected cells. Here is described the solution to design an optical system, which allow backscattering Mie diffusion spectra determination for biological sample on transparent holder. Fractal dimension as interest area of Mie scattering spectra was computed using a software package, which can store and process recorded data. The results indicate obviously higher values of fractal dimension for HSV-infected biological samples when compared with non-infected biological samples. This allows us to clearly discriminate between virus-infected and non-infected biological samples.

Large Scale Wireless Sensor Networks Based on Fixed Nodes and Mobile Robots in Precision Agriculture

The paper presents an innovative design for wireless sensor networks (WSNs) which can increase their performances in the precision agriculture (PA). The authors argue on the reasons why presently WSNs are not integrated in PA on a larger scale. To this end, the paper describes the state of art in WSNs used in PA, and develops the proposed solution on the basis of the identified conjectures. The conceived solution proves how nodes can have small dimensions without mitigating the communication range or energy autonomy. Small dimensions bring benefits also to costs and operations. Such results can be achieved by overcoming the misalignment between advances in unmanned agricultural vehicles (AVs) and advances in WSN. The fixed nodes and mobile nodes (AVs), cooperate to fulfil sensor and communication coverage. Endowing the WSN nodes only with necessary sensors together with an energy awareness communication algorithm adopted in resonance with PA requirements proves to be a new competitive approach. Some simulations to validate these concepts are also provided.

Emotion Classification Using a Tensorflow Generative Adversarial Network Implementation

The detection of human emotions has applicability in various domains such as assisted living, health monitoring, domestic appliance control, crowd behavior tracking real time, and emotional security. The paper proposes a new system for emotion classification based on a generative adversarial network (GAN) classifier. The generative adversarial networks have been widely used for generating realistic images, but the classification capabilities have been vaguely exploited. One of the main advantages is that by using the generator, we can extend our testing dataset and add more variety to each of the seven emotion classes we try to identify. Thus, the novelty of our study consists in increasing the number of classes from N to 2N (in the learning phase) by considering real and fake emotions. Facial key points are obtained from real and generated facial images, and vectors connecting them with the facial center of gravity are used by the discriminator to classify the image as one of the 14 classes of interest (real and fake for seven emotions). As another contribution, real images from different emotional classes are used in the generation process unlike the classical GAN approach which generates images from simple noise arrays. By using the proposed method, our system can classify emotions in facial images regardless of gender, race, ethnicity, age and face rotation. An accuracy of 75.2% was obtained on 7000 real images (14,000, also considering the generated images) from multiple combined facial datasets.

Integrating UAV in IoT for RoI Classification in Remote Images

The paper presents a cheap and efficient solution for remote processing of images taken by a team of UAVs (Unmanned Aerial Vehicles). The work objective was to implement an integrated system for detection and classification of regions of interest (RoIs), in the case of flood events. The UAVs are considered as objects of the internet. This means the integration of UAVs in IoT (Internet of Things) as intelligent objects. The investigated RoIs are: water, grass, forests, buildings, and roads. A multi UAV – multi GCS (Ground Control Station) solution is proposed. Due to this integration, land segmentation by image processing can be efficiently made in real time. For RoI detection and evaluation a multi CNN structure is used as a multi classifier structure. Particularly, a CNN classifier is implemented for each type of RoI and all the CNNs work in parallel. The orthophotoplan obtained from remote acquired images are successively decomposed in adjacent images of size 6000 × 4000 and next in overlapping patches of size 65 × 65 pixels for classifier learning or for testing. Finally, the images are segmented in RoIs by a multi-mask technique and the percentage of each RoI is calculated. The accuracy of segmentation and the processing time, evaluated from 10 real images, was better than in other reported cases.