Yasir Salih

Comparison of stochastic filtering methods for 3D tracking

In the recent years, the 3D visual research has gained momentum with publications appearing for all aspects of 3D including visual tracking. This paper presents a review of the literature published for 3D visual tracking over the past five years. The work particularly focuses on stochastic filtering techniques such as particle filter and Kalman filter. These two filters are extensively used for tracking due to their ability to consider uncertainties in the estimation. The improvement in computational power of computers and increasing interest in robust tracking algorithms lead to increase in the use of stochastic filters in visual tracking in general and 3D visual tracking in particular. Stochastic filters are used for numerous applications in the literature such as robot navigation, computer games and behavior analysis. Kalman filter is a linear estimator which approximates systems dynamics with Gaussian model while particle filter approximates systems dynamics using weighted samples. In this paper, we investigate the implementation of Kalman and particle filters in the published work and we provide comparison between these techniques qualitatively as well as quantitatively. The quantitative analysis is in terms of computational time and accuracy. The quantitative analysis has been implemented using four parameters of the tracked object which are object position, velocity, size of bounding ellipse and orientation angle.

Tone mapping of HDR images: A review

Real world contains a wide range of intensities that cannot be captured with traditional imaging devices. Moreover, even if these images are captured with special procedures, existing display devices cannot display them. This paper presents a comparative study of most famous tone mapping algorithms. Tone mapping is the process of compressing high dynamic range images into a low dynamic range so they can be displayed by traditional display devices. The study implements six tone mapping algorithms and performs a comparison between them by visual rating. Independent participant were asked to rate these images based on a given rating scheme. The study concluded that Reinhard tone mapping operators are the best in term of visual pleasure and maintaining image integrity. In addition, exponential tone mapping operators have achieved better rating compared the logarithmic operators.

An intelligent control of Blood Pressure system using PID and Neural Network

This paper presents an intelligent control approach for Blood Pressure (BP) system using PID-Neural Network (PIDNN). This paper discusses the use of PIDNN to optimize the regulation of Mean Arterial Pressure (MAP) through the infusion of Sodium Nitroprusside (SNP). Multi-layer Perception (MLP) is used as Neural Network pattern predictor. Simulation studies based on a sensitive model of MAP shows that the proposed approach is able provide a good performance.

3D Tracking using particle filters

Recently, Particle filter has been used for numerous 3D tracking applications especially nonlinear tracking applications which are intractable using Kalman filter or other linear estimator. Particle filter approximates systems dynamics using weighted samples; therefore it can work with variety of systems. In the literature, particle filter is mostly used for articulated body tracking, gesture recognition and robot tracking. Although other applications exist, these are the dominant ones. This paper discusses 3D object tracking using particle filters. Three main particle filtering algorithms have been discussed in this paper and their performances have been evaluated using RMSE performance measure.

Depth and Geometry from a Single 2D Image Using Triangulation

We present a novel method for computing depth of field and geometry from a single 2D image. This technique, unlike the existing ones measures the absolute depth of field and distances in the scene from single image only using the concept of triangulation. This algorithm requires minimum inputs such as camera height, camera pitch angle and camera field of view for computing the depth of field and 3D coordinates of any given point in the image. In addition, this method can be used to compute the actual size of an object in the scene (width and height) as well as the distance between different objects in the image. The proposed methodology has the potential to be implemented in high impact applications such as distance measurement from mobile phones, robot navigation and aerial surveillances.

An algorithm for vehicle detection and tracking

In this paper, we propose a vehicle detection and tracking algorithm. The detection is done using the median filtering and blob extraction. Median filtering is used for background extraction which is later subtracted from the motion frames for object detection. Morphological operators are employed for blob extraction. Hence, object detection is achieved using median filtering and morphological closing operation. Kalman filtering is used for object tracking which uses location of blobs. One of the advantages of this system is that each vehicle in the frame is classified into different color boxes. We present preliminary research results that will finally lead to the identification of the tracked vehicle.

Digital assessment of facial acne vulgaris

Acne affects 85% of adolescents at some time during their lives. Dermatologists use manual methods such as direct visual assessment and ordinary flash photography to assess the acne. However, these manual methods are time consuming and may result in intra-observer and inter-observer variations, even by experienced dermatologists. The objective of this research is to develop a computational imaging method for automated acne grading. The first step in the proposed method is pre-processing which involves lighting compensation. The CIE La*b* color space is used to measure any dissimilarity between skin colors. Acne segmentation has been performed using automated modified K-means clustering algorithm and support vector machines (SVM) classifier. Color and diameter are the main features extracted to classify acne blobs into different acne classes; papule, pustule, nodule or cyst. Finally, the severity level is determined such as mild, moderate, severe and very severe.

Segmentation assessment of activated sludge flocs at different magnifications for wastewater treatment

Activated sludge process form an important part of wastewater treatment plant with domestic effluent. The variations in the state of the process are appeared as those in the size and structure of flocs and filaments found in the wastewater samples from aeration tank of secondary treatment. The normal operation requires proper settling of flocs in the secondary clarifier, which is affected by problem of bulking and pin point flocs. Conventional physico-chemical methods take a lot of time to detect the abnormal operation, consequently leaving insufficient time for precautionary measures. Image processing and analysis of microscopic images can offer a time-efficient alternative to monitor the operation of activated sludge process. Segmentation is a necessary part of image processing and analysis for identification of regions of interest in the image, and its acceptable accuracy is pre-requisite of the morphological analysis. In this paper, three segmentation techniques, fuzzy cmeans, k-means and Otsu thresholding, were used to segment flocs in microscopic images of samples taken from aeration tank of activated sludge process. The performance of the segmentation algorithms was evaluated for images taken at four different objective magnifications of microscope, using metrics of global consistency error (GCE), random index (RI) and variation of information (VI). The performance metrics were evaluated by comparing the segmented images with the approximation of ground truth images. Finally, the effect of magnification was investigated on the image segmentation and analysis procedure and observed that the size of floc, perceptible to the image segmentation and analysis procedure is greater and more precise at higher magnification.

3D object tracking using three Kalman filters

In the recent years, 3D tracking has gained attention due to the perforation of powerful computers and the increasing interest in tracking applications. One of the most common tracking algorithms used is the Kalman filter. Kalman filter is a linear estimator that is based on approximating systems dynamics using Gaussian probability distribution. In this paper, we provide a detailed evaluation of the most common Kalman filters, their use in the literature and their implementation for 3D visual tracking. The main types of Kalman filters discussed are linear Kalman filter, extended Kalman filer and unscented Kalman filter.

Stochastic filters for object tracking

Stochastic filters have been extensively used for object tracking because of its ability to measure uncertainties and high accuracy. In recent years, the availability of cheap computers with high computational power has led to incorporate tracking systems in many consumer electronics devices such as surveillance cameras and game consoles. In this paper, we compare Kalman filter and particle filter tracking based on their computational time and estimation accuracy. These two filters represent 50% of the published work on object tracking in the last five years.