Search

In recent years, movement related cortical potentials (MRCP), a type of slow cortical potentials, have been used for motor intention detection for triggering external devices in close loop rehabilitation paradigm. One of the main issues with these slow frequency MRCP signals is to separate them from the background brain activity or their poor signal to noise ratio. For this reason, different filter bank for denoising are reported in the literature. In this paper, we use a signal adaptive, time-frequency based empirical method known as empirical mode decomposition to decompose the MRCP signal into its constituent frequency components called intrinsic mode function (IMF). By doing so, we can separate the slow frequency MRCP signal occupying the last few IMFs, from the high frequency noise present in the first few IMFs. By removing the first few IMFs, we obtain a cleaned MRCP signal that can be used for the detection of movement intention with much higher accuracies and less latency. Template matching approach of matched filters was used to detect the presence of signal and the results show that four different types of movement intentions can be detected using this denoising approach with an average latency of 0.70 s±0.1 before the actual movement onset. These results are much better than average latency of 0.07 s ± 0.1 reported previously in the literature on the same dataset.

Presence of neurobiological disorder, like epilepsy, causes abnormalities in brain functionality. This is why connectivity and activity patterns of brain regions in epileptic patients are very different as compared with healthy subjects. These asymmetries can be used to distinguish epileptic patients from healthy subjects. Robust features are extracted to capture asymmetries in connectivity patterns. However, these features do not give any information about the brain activity in a particular local region. So in order to get better classification accuracy, we need robust features to capture asymmetries in regional activities. In this paper, we propose a novel feature set that captures abnormalities in activities at local regional level by finding inter subject blood oxygen level dependent (BOLD) signal asymmetries (iBSA). By combining this iBSA with asymmetries in global connectivity patterns, we are able to capture all asymmetries at global network as well as local regions in only 225 features. Employing these 225 features and support vector machines for classification, an overall accuracy of 86.6% was obtained on resting state fMRI (rfMRI) data of 180 subjects without any feature selection. The results presented in this paper are better than any other reported results in current literature.

Human detection has gained considerable importance in aggravated security scenarios over recent times. An effective security application relies strongly on detailed information regarding the scene under consideration. A larger accumulation of humans than the number of personal authorized to visit a security controlled area must be effectively detected, amicably alarmed and immediately monitored. A framework involving a novel combination of some existing techniques allows an immediate detection of an undesirable crowd in a region under observation. Frame differencing provides a clear visibility of moving objects while highlighting those objects in each frame acquired by a real time camera. Training of a correlation pattern recognition based filter on desired shapes such as elliptical representations of human faces (variants of an Omega Shape) yields correct detections. The inherent ability of correlation pattern recognition filters caters for angular rotations in the target object and renders decision regarding the existence of the number of persons exceeding an allowed figure in the monitored area.

Consumer electronic devices require brightness preservation while image enhancement. Many of the histogram equalization techniques that have been introduced tend to strike out the mean image brightness while the process of enhancement is carried out. Hence, we suggest a method that enhances the image while keeping its brightness in a better form as compared to the former methods applied. Firstly, based on median value, the input image histogram is divided into sub-histogram and this process reprises until we have derived 8 sub-histograms. Further, the clipping of the histogram is implemented based on the average pixels in each sub-histogram. Then, each sub-histogram is assigned with a new dynamic range and finally customary histogram equalization is done independently on each sub-histogram. Proposed method is tested against conventional methods and results indicate that image mean brightness is preserved better than other methods while enhancing the image. The proposed method keep the brightness save of the original image while during the enhancement of the images. Many images has been taken from the SIPI database and applied the proposed method for the contrast enhancement of the low contrast images and results are improved. The major problem is to keep the brightness of the low contrast image preserve. The proposed technique has dealt with brightness preservation of the images.

A novel rotation and scale invariant texture classification methodology is proposed based on distribution matching in higher dimensional space. Feature extraction is performed by using uniform local binary patterns (uLBPs) in which the rotation and scale changes in an image cause shifts in the underlying uLBP histograms. To compensate for these shifts at the classification layer, the distributions of training and testing data using kernel methods are estimated and means of the two distributions in the transformed domain using importance weights are matched. These calculated importance weights are used in the standard support vector machines to compensate for the shift in the distributions. The proposed method is used for classifying the images in the Brodatz texture database demonstrating the effectiveness of the proposed methodology.

The traditional formulation of segmentation of images using active contours with edges exhibits an inherent shortcoming: when the boundaries between skin and lesions is smooth, edge based features tend to give poor segmentation results. An alternative formulation of active contours using structure tensor is proposed in which creasness features (peaks and valleys) are used for the determination of lesion boundaries, which enhance directional gradients while inhibiting local texture attributes in the images. The creasness features are used for formulation of external energy term in the active contour framework to segment dermoscopy images. Experiments show that the proposed segmentation method exhibits good performance comparing favorably to the traditional edge based active contour formulation.

The introduction of various novel imaging technologies such as narrow-band imaging have posed novel image processing challenges to the design of computer assisted decision systems. In this paper, we propose an image descriptor referred to as integrated scale histogram local binary patterns. We propagate an aggregated histogram of local binary patterns of an image at various resolutions. This results in low dimensional feature vectors for the images while incorporating their multiresolution analysis. The descriptor was used to classify gastroenterology images into four distinct groups. Results produced by the proposed descriptor exhibit around 92% accuracy for classification of gastroenteroloy images outperforming other state-of-the-art methods, endorsing the effectiveness of the proposed descriptor.

Segmentation is a vital step for pattern recognition systems used in in-body imaging scenarios. In this paper we compare the performance of three popular segmentation algorithms (mean shift, normalized cuts, level-sets) when applied to two distinct in-body imaging scenarios: chromoen-doscopy and narrow-band imaging. Observation shows that the model-based algorithm did not perform well, when compared to its segmentation by clustering alternatives. Normalized cuts obtained the best performance although future work hints that texture similarity should be further explored in order to increase segmentation performance in this type of scenarios.

Automatic classification of cancer lesions for gastroenterology imaging scenarios poses novel challenges to computer assisted decision systems, owing to their distinct visual characteristics such as reduced color spaces or natural organic textures. In this paper, we explore the prospects of using Gabor filters in a texton framework for the classification of images from two distinct imaging modalities (chromoendoscopy and narrow-band imaging) into three different groups: normal, precancerous and cancerous. Results show that they produce consistent results for both imaging modalities, hinting on their possible generic use for the classification of in-body images.

To overcome the lack of strict QoS guarantees in existing hotspots, in this paper we present a system for realtime traffic support in 802.11 networks that works in either infrastructure or ad-hoc mode. The proposed mechanism, called elastic MAC (E-MAC) protocol, helps stations with real-time traffic to organize and establish a transmission schedule in a distributed manner, while coexisting with standard 802.11 stations. This distributed scheduling guarantees very short delays and a minimum reserved data rate for real-time stations, while protecting best-effort 802.11 traffic from starvation. Another feature of our mechanism is time-slot reuse, which improves the network efficiency by allowing other real-time stations to take over unused slots (e.g., in case of using voice codecs with silence suppression). We evaluate the performance of our system using a testbed implementation, ns-2 simulations and a mathematical model, and show how it outperforms other QoS schemes (e.g., 802.11e) in terms of throughput, delay, and jitter.

Automatic classification of cancer lesions in tissues observed using gastroenterology imaging is a non-trivial pattern recognition task involving filtering, segmentation, feature extraction and classification. In this paper we measure the impact of a variety of segmentation algorithms (mean shift, normalized cuts, level-sets) on the automatic classification performance of gastric tissue into three classes: cancerous, pre-cancerous and normal. Classification uses a combination of color (hue-saturation histograms) and texture (local binary patterns) features, applied to two distinct imaging modalities: chromoendoscopy and narrow-band imaging. Results show that mean-shift obtains an interesting performance for both scenarios producing low classification degradations (6%), full image classification is highly inaccurate reinforcing the importance of segmentation research for Gastroenterology, and confirm that Patch Index is an interesting measure of the classification potential of small to medium segmented regions.

In this paper, we present a numerical comparison of how well segmentation algorithms approximate the manual segmentation of gastroenterologists for a set of endoscopic images. Different areas in these images demand different levels of analysis by a clinician and some provide critical information about the patient. Our objective is thus to segment endoscopic images so that the results mimic as closely as possible the areas that were considered relevant by doctors. We focus on a detailed quantitative comparison of two popular segmentation algorithms, mean shift and normalized cuts, when applied to in-body images, most specifically for vital-stained magnification endoscopy. Segmentation results are compared with the manual annotations of the same images performed by two specialist clinicians. Results show that if we simply consider the most relevant segmented patch, normalized cuts performs better. However, if we allow the annotated area to be represented by multiple patches, mean shift is clearly a better choice, although automatic ways to determine its kernel's bandwidth are highly desirable.

The design of computer-assisted decision (CAD) systems for different biomedical imaging scenarios is a challenging task in computer vision. Sometimes, this challenge can be attributed to the image acquisition mechanisms since the lack of control on the cameras can create different visualizations of the same imaging site under different rotation, scaling, and illumination parameters, with a requirement to get a consistent diagnosis by the CAD systems. Moreover, the images acquired from different sites have specific colors, making the use of standard color spaces highly redundant. In this paper, we propose to tackle these issues by introducing novel region-based texture, and color descriptors. The proposed texture features are based on the usage of analytic Gabor filters (for compensation of illumination variations) followed by the calculation of first- and second-order statistics of the filter responses and making them invariant using some trivial mathematical operators. The proposed color features are obtained by compensating for the illumination variations in the images using homomorphic filtering followed by a bag-of-words approach to obtain the most typical colors in the images. The proposed features are used for the identification of cancer in images from two distinct imaging modalities, i.e., gastroenterology and dermoscopy . Experiments demonstrate that the proposed descriptors compares favorably to several other state-of-the-art methods, elucidating on the effectiveness of adapted features for image characterization.