Y. Roodt’s research while affiliated with University of Johannesburg and other places

What is this page?


This page lists works of an author who doesn't have a ResearchGate profile or hasn't added the works to their profile yet. It is automatically generated from public (personal) data to further our legitimate goal of comprehensive and accurate scientific recordkeeping. If you are this author and want this page removed, please let us know.

Publications (8)


Blind deconvolution of Gaussian blurred images containing additive white Gaussian noise
  • Conference Paper

February 2013

·

43 Reads

·

7 Citations

·

Y. Roodt

Image restoration algorithms are used to reconstruct the information that is suppressed when an observed image is subjected to blurring. These algorithms generally assume that knowledge of the nature of the distortion and noise contained in an observed image is available. When this information is not available and has to be directly estimated from the image being processed the problem becomes one of blind deconvolution. This paper makes use of a novel blur identification technique and a noise identification technique to perform blind deconvolution on single images that have been degraded by a Gaussian blur and contain additive white Gaussian noise.


Robust single image noise estimation from approximate local statistics
  • Conference Paper
  • Full-text available

September 2012

·

12 Reads

·

3 Citations

Yuko Roodt

·

·

·

[...]

·

Y. Roodt

A novel method for estimating the variance and standard deviation of the additive white Gaussian noise contained in an image will be presented. Only a single image is used to estimate the noise properties. Local image outliers are discarded, this allows us to separate the additive zero mean white Gaussian noise contained in a noisy image from the original image structure. Local variance estimates can then be calculated from the extracted noise. These local variance estimates are weak and can be influenced by misclassified image information. Robust statistics are then used to fuse the weak local variance estimates to obtain a robust global noise variance estimate. This method of estimating the noise properties is computationally efficient and provides reliable estimation results in synthetic and real-world imagery. The accuracy and processing complexity of the proposed algorithm will be compared against the current state-of-the-art noise estimators.

Download

Figure 2: Diagrammatical view of the classical CPU and GPU setup (adapted from Owen et al. 2007)
Figure 3: The Conjugate Gradient method
Figure 4: Parallel reduction, used on the GPU
Figure 6: Performance of a single iteration of the Conjugate Gradient
Specifications of the different computers
The Potential of Graphical Processing Units to Solve Hydraulic Network Equations

July 2012

·

931 Reads

·

21 Citations

Journal of Hydroinformatics

The Engineering discipline has relied on computers to perform numerical calculations in many of its sub-disciplines over the last decades. The advent of graphical processing units (GPUs), parallel stream processors, has the potential to speed up generic simulations that facilitate engineering applications aside from traditional computer graphics applications, using GPGPU (general purpose programming on the GPU). The potential benefits of exploiting the GPU for general purpose computation require the program to be highly arithmetic intensive and also data independent. This paper looks at the specific application of the Conjugate Gradient method used in hydraulic network solvers on the GPU and compares the results to conventional central processing unit (CPU) implementations. The results indicate that the GPU becomes more efficient as the data set size increases. However, with the current hardware and the implementation of the Conjugate Gradient algorithm, the application of stream processing to hydraulic network solvers is only faster and more efficient for exceptionally large water distribution models, which are seldom found in practice.


Automated surveillance and detection of foreign stationary objects

September 2011

·

12 Reads

·

3 Citations

CCTV systems are frequently monitored manually by a human observer. This human observer is typically responsible for dealing with tens or hundreds of cameras at a time. Potential security threats may easily be missed by the system's human operators due to fatigue or being overwhelmed by the amount of change in the images. The timely detection of security threats is an important attribute for any security system. A robust algorithm for detecting potential threats from a surveillance video is presented.


Stationary region predictor using a stationary camera

January 2011

·

1 Read

A method to determine the stationery probability of regions or feature points in a video sequence is proposed in this paper. This is done by identifying feature points using the Harris corner detector, finding descriptors for the feature points and then tracking the feature points. The information gained from tracking the feature points is then used to determine the stationery probability of these features. This method is shown to successfully identify probable stationery and moving regions in video sequences.


Image colourisation for compression using GPU hardware

There is a growing demand for high definition (HD) graphics with multimedia content. This demand requires significantly more computational power than before. The increased demand in video content will continue to grow, resulting in vast volumes of data continuously shifted across networks and the internet. The volume of video data must be decreased in order to better align to trends while maintaining efficient transmission and real-time processing requirements. In this work, the volume of information in HD images are decreased by reducing the image to greyscale while maintaining real-time performance requirements. The real-time processing requirement is met by shifting the computation to the graphics processing unit (GPU). The GPU is a programmable massively parallel processing unit with numerous cores. The performance of the colourisation technique which adds chrominance to the greyscale image is drastically increased using the GPU, which would otherwise take several minutes to colourise a 1080p image. The results indicate that a video compression scheme for HD video is viable.


Image processing on the GPU: Implementing the Canny edge detection algorithm

November 2007

·

10 Reads

·

18 Citations

In this paper we present a detailed Graphics Processing Unit (GPU)-based implementation of the well known Canny edge detection algorithm. The aim of the paper is to provide an overview on our approach to implement the Canny edge detection algorithm, as it encompasses a set of image processing techniques. The result is an algorithm that can be applied in real-time applications. A basic understanding of the hardware architecture and available tools are required to successfully map computer vision and image processing techniques to the GPU. We describe some of the benefits associated with this platform while looking at possible development pitfalls, solutions and performance results.


Fig. 1. 
Fig. 2. 
Fig. 3. 
Text segmentation and recognition in unconstrained imagery

In this paper, we present a novel method for rec-ognizing and segmenting symbols and text in complex image sequences. The algorithm is designed to take advantage of the massive computing capability of parallel processing architectures. The additional processing resources will allow for more prepro-cessing steps reducing the number of simplification assumptions on the orientation, structure, scale and colour of the detected character symbols. The increased algorithmic complexity yields better recognition performance. This optical character recog-nition framework was designed to run on video sequences of unstructured environments. A robust algorithm will be presented that addresses these underlining vision based issues and will be tested for speed and recognition accuracy.

Citations (5)


... Then the final version of the video was run in a loop to perfectly see and track moving objects in the video. Moreover, it can also separate the preferred objects from these videos by further applying thresholds to present data (Roodt et al., 2007). 6) Motion Detection: ...

Reference:

A parallel computing framework for real-time moving object detection on high resolution videos
Image processing on the GPU: Implementing the Canny edge detection algorithm

... The mean value was 0.48979534, with a standard deviation of 0.00032867. Because the object is a homogeneous phantom, the calculation with a lower standard deviation is better [1,[22][23][24][25]. ...

Robust single image noise estimation from approximate local statistics

... Additionally, several studies have explored the use of hardware devices to enhance the computational efficiency of large-scale water distribution networks. For example, Corus et al. investigated the possibility of using GPU parallel computing technology to solve the conjugate gradient method [22]. By altering the computational formula of the conjugate gradient method to avoid matrix inversion, they proposed a parallel method that could run on graphics processors for numerically solving hydraulic models of distribution networks. ...

The Potential of Graphical Processing Units to Solve Hydraulic Network Equations

Journal of Hydroinformatics

... function is used to generate the Gaussian noise. In this study, we applied Gaussian blur by varying the blur factor from 1 to 10, with the levels being 1, 2, 5, 8, and 10, as represented by Equation (4), where I is the original image without the blur, I blurred is the blurred image that contains Gaussian blur, G is a Gaussian kernel with standard deviation σ, and ⊗ denotes the convolution operation [41]. Gaussian blur is achieved by convolving the image with a Gaussian kernel using a function called 'ImageFilter.GaussianBlur()'. ...

Blind deconvolution of Gaussian blurred images containing additive white Gaussian noise
  • Citing Conference Paper
  • February 2013

... Modern surveillance networks have many industrial applications including monitoring patients in hospitals, detecting violence in stadiums, and identifying lost baggage in airports (Kiewiet et al., 2011;Ramachandran et al., 2012;Wang, 2013). Regardless of application, these systems have the same underlying goals: to detect and report objects fields of computer vision and expert systems (Andreopoulos & Tsotsos, 2013) and have been used extensively in automated surveillance systems (Kiewiet et al., 2011;Ramachandran et al., 2012;Shah et al., 2007). ...

Automated surveillance and detection of foreign stationary objects
  • Citing Conference Paper
  • September 2011