OpenUCT :: Browsing by Author "De Jager, Gerhard"

Browsing by Author "De Jager, Gerhard"

Now showing 1 - 20 of 45

Open Access
3D model reconstruction using photoconsistency
(2007) Joubert, Kirk Michael; Nicolls, Fred; De Jager, Gerhard
Model reconstruction using photoconsistency refers to a method that creates a photohull, an approximate computer model, using multiple calibrated camera views of an object. The term photoconsistency refers to the concept that is used to calculate the photohull from the camera views. A computer model surface is considered photoconsistent if the appearance of that surface agrees with the appearance of the surface of the real world object from all camera viewpoints. This thesis presents the work done in implementing some concepts and approaches described in the literature.
Open Access
3D reconstruction and camera calibration from 2D images
(2000) Henrichsen, Arne; De Jager, Gerhard
A 3D reconstruction technique from stereo images is presented that needs minimal intervention from the user. The reconstruction problem consists of three steps, each of which is equivalent to the estimation of a specific geometry group. The first step is the estimation of the epipolar geometry that exists between the stereo image pair, a process involving feature matching in both images. The second step estimates the affine geometry, a process of finding a special plane in projective space by means of vanishing points. Camera calibration forms part of the third step in obtaining the metric geometry, from which it is possible to obtain a 3D model of the scene. The advantage of this system is that the stereo images do not need to be calibrated in order to obtain a reconstruction. Results for both the camera calibration and reconstruction are presented to verify that it is possible to obtain a 3D model directly from features in the images.
Open Access
Application of the watershed boundary technique to automatically segment surface froth images
(1995) Liu, Jia; De Jager, Gerhard
The purpose of this dissertation is to investigate the suitability of using a recently proposed computer processing algorithm - the watershed boundary technique for applications in computer vision systems, where on-line segmentation of the surface froth images in commercial flotation cells is required. In industrial flotation cells, the surface froth offers considerable visual information as to the grade and recovery in extraction and the concentration of minerals in raw ores. The individual bubbles that constitute the surface froth give rise to complex three-dimensional structures that are problematic to segment accurately and reliably by computer vision. Investigating an efficient technique to automatically and accurately extract these visual features in real time is therefore the main task of this research work. Past research work into this field has resulted in a number of different techniques and algorithms for the purpose of segmentation. However, these algorithms are often iterative and therefore quite slow. The technique described here is noniterative and therefore one with industrial real time processing implications. The results show that the watershed boundary technique provides a reliable method for the segmentation of surface froth structures. Minor errors which occur do not significantly influence the statistical parameters which can be determined from the segmented images. This technique is not only very successful in segmentation, but may also be implemented for real-time operation. This clearly leads to a viable new segmentation method.
Open Access
Automatic detection and segmentation of brain lesions from 3D MR and CT images
(2014) Mokhomo, Molise; Nicolls, Fred; De Jager, Gerhard; Muller, N
The detection and segmentation of brain pathologies in medical images is a vital step which helps radiologists to diagnose a variety of brain abnormalities and set up a suitable treatment. A number of institutes such as iThemba LABS still rely on a manual identification of abnormalities. A manual identification is labour intensive and tedious due to the large amount of medical data to be processed and the presence of small lesions. This thesis discusses the possible methods that can be used to address the problem of brain abnormality segmentation in MR and CT images. The methods are general enough to segment different types of abnormalities. The first method is based on the symmetry of the brain while the second method is based on a brain atlas. The symmetry-based method assumes that healthy brain tissues are symmetrical in nature while abnormal tissues are asymmetric with respect to the symmetry plane dividing the brain into similar hemispheres. The three major steps involved in this approach are the symmetry detection, tilt correction and asymmetry quantification. The method used to determine the brain symmetry automatically is discussed and its accuracy has been validated against the ground-truth using mean angular error (MAE) and distance error (DE). Two asymmetric quantification methods are studied and validated on real and simulated patient’s T1- and T2-weighted MR images with low and highgrade gliomas using true positive volume fraction (TPVF), false positive volume fraction (FPVF) and false negative volume fraction (FNVF). The atlas-based method is also presented and relies on the assumption that abnormal brain tissues appear with intensity values different from those of the surrounding healthy tissues. To detect and segment brain lesions the test image is aligned onto the atlas space and voxel by voxel analysis is performed between the atlas and the registered image. This methods is also evaluated on the simulated T1-weighted patient dataset with simulated low and high grade gliomas. The atlas, containing prior knowledge of normal brain tissues, is built from a set of healthy subjects.
Open Access
The automatic detection of patterns in people's movements
(2002) Forbes, Gordon; De Jager, Gerhard
Bibliography: leaves 102-105.
Open Access
Calibration, recognition, and shape from silhouettes of stones
(2007) Forbes, Keith; Nicolls, Fred; De Jager, Gerhard
Multi-view shape-from-silhouette systems are increasingly used for analysing stones. This thesis presents methods to estimate stone shape and to recognise individual stones from their silhouettes. Calibration of two image capture setups is investigated. First, a setup consisting of two mirrors and a camera is introduced. Pose and camera internal parameters are inferred from silhouettes alone. Second. the conﬁguration and calibration of a high throughput multi-camera setup is covered. Multiple silhouette sets of a stone are merged into a single set by inferring relative poses between sets. This is achieved by adjusting pose parameters to maximise geometrical consistency speciﬁed by the epipolar tangency constraint. Shape properties (such as volume, flatness, and eiongation) are inferred more accurately from the merged silhouette sets than from the original silhouette sets. Merging is used to recognise individual stones from pairs of silhouette sets captured on different occasions. Merged sets with sufﬁcient geometrical consistency are classiﬁed as matches (produced by the same stone), whereas inconsistent sets are classiﬁed as mismatches. Batch matching is determining the one-to-one correspondence between two unordered batches of silhouette sets of the same batch of stones. A probabilistic framework is used to combine recognition by merging (which is slow, but accurate) with the efﬁciency of computing shape distribution-based dissimilarity values. Two unordered batches of 1200 six-view silhouette sets of uncut gemstones are correctly matched in approximately 68 seconds (using a 3.2 GHz Pentium 4 machine]. An experiment that compares silhouette-based shape estimates with mechanical sieving demonstrates an application using the developed methods. A batch of 494 garnets is sieved 15 times. After each sieving, silhouette sets are captured for sub-batches in each bin. Batch matching is used to determine the IS sieve bins per stone. Better estimates of repeatability, and better understanding of the variability of the sieving process is obtained than if only histograms (the natural output of sieving) were considered. Silhouette-based sieve emulation is found to be more repeatable than mechanical sieving.
Open Access
Classification of cured tobacco leaves by colour and plant position by means of computer processing of digital images
(1999) Tattersfield, George Metcalf; De Jager, Gerhard
This dissertation investigates the machine vision grading of flue-cured Virginia tobacco by means of digital processing of tobacco leaf images. With reference to international grading standards and to modem image processing techniques, two classifiers are designed. The colour classifier uses seven features extracted from each leaf image to grade the leaf into one of five official colour classes. It does this with an expected correct classification rate of 93.5%. The plant position classifier identifies the position on the stalk from which a leaf was reaped, using ten size and shape features to classify the leaf into one of six plant position categories. It has a correct classification rate of 70%. Average colours for each colour class and archetypal shapes for each plant position category are derived from the digital leaf data. These should be of value to tobacco graders as objective representations of typical leaves within each class.
Open Access
Colour analysis and the classification of fruit
(1992) Kay, Gary R; De Jager, Gerhard
The increasing high standards of fruit quality expected by the agricultural export market of South Africa has reached a stage that fruit must be accurately graded in a short a time as possible. This thesis describes colour systems and methods to grade the fruit automatically via the clustering and classification methods. After investigating several approaches to automatically sort fruit based on colour, an image processing approach was taken. The colours on the fruit (specifically apples) were analyzed, by capturing a colour image of the fruit and analyzing the pixels in the image. Several colour representation systems were investigated and they are: colours represented by spectral power distributions and spectral reflectance curves; the CIE 1931 XYZ tristimulus values; the CIE 1931 x ,y ,z chromaticity coordinates; the CIE 1960 L, u, v uniform chromaticity scale (UCS); the Munsell colour wheel of hue, value and chroma (HVC); the L*u*v* system; the L *a*b* system; the Red, Green and Blue (RGB) system; and the hue, saturation and intensity (HSI) perceptual colour representations. In addition, several clustering and classification techniques were investigated and they are: the supervised methods of Parametric Bayesian classification and minimum Euclidean distance classification; and the unsupervised methods of the K-means algorithm and the ISODATA classification approach. The ICS Texicon computer spectrophotometer (ICS Texicon Spectraflash Manual (1991)) was used to check the performance of most of the colour systems described by analyzing apple sample colours
Open Access
Computer aided diagnosis of miliary TB in chest X-rays
(2001) Koeslag, Anthony; De Jager, Gerhard
With the improvement in computer technology, Computer Aided Diagnosis (CAD) is becoming an increasingly more powerful tool for radiologists. The focus of this project was on CAD of pulmonary miliary tuberculosis. Several methods for enhancing lung textures were discussed as an aid to the radiologist in diagnosing miliary TB. Some statistical approaches and template matching methods were used to measure characteristics of both healthy and unhealthy (miliary TB) lung textures. These measurements were evaluated to see if a computer can be programmed to differentiate between lung texture from a healthy lung and lung texture from a lung with miliary TB.
Open Access
The design, implementation and analysis of a wavelet-based video codec
(1998) Servais, Marc Paul; De Jager, Gerhard
The Wavelet Transform has been shown to be highly effective in image coding applications. This thesis describes the development of a new wavelet-based video compression algorithm which is based on the 3D wavelet transform, and requires no complicated motion estimation techniques. The proposed codec processes a sequence of images in a group of frames (GOF) by first transforming the group spatially and temporally, in order to obtain a GOF of 3D approximation and detail coefficients. The codec uses selective prediction of temporal approximation coefficients in order to decorrelate transformed GOFs. Following this, a modified version of Said and Pearlman's image coding technique of Set Partitioning in Hierarchical Trees is used as a method for encoding the transformed GOF. The compression algorithm has been implemented in software, and tested on seven test sequences at different bit-rates. Experimental results indicate a significantly improved performance over MPEG 1 and 2 in terms of picture quality, for sequences filmed with a stationary camera. The codec also performs well on scenes filmed with a moving camera, provided that there is not a large degree of spatial detail present. In addition, the proposed codec has several attractive features. It performs well without entropy coding, and does not require any computationally-expensive motion estimation methods, such as those used by MPEG. Finally, a substantial advantage is that the encoder generates a bit-stream which allows for the progressive transmission of video, making it well-suited to use in video applications over digital networks.
Open Access
Designing hypothesis tests for digital image matching
(2000) Cox, Gregory Sean; Wohlberg, Brendt; Nicolls, Fred; De Jager, Gerhard
Image matching in its simplest form is a two class decision problem. Based on the evidence in two sensed images, a matching procedure must decide whether they represent two views of the same scene, or views of two different scens. Previous solutions to this problem were either based on an intuitive notion of image similarity, or were modelled on solutions to the superficially similar problem of target detection in images. This research, in contrast, uses a decision theoretic formulation of the problem, with the image pair as unit of observation and probability of error in the match/mismatch decision as performance criterion. A stochastic model is proposed for the image pair, and the optimal test of match and mismatch hypotheses for samples of this random process is derived. The test is written conveniently in terms of a statistic of the two images and a scalar decision threshold. The analytical advantages of a solution derived from first principles are illustrated with the derivation of hypothesis conditional probability distributions, optimal decision thresholds, and expessions for the probability of error in the decision.
Open Access
The development of a predictive autofocus algorithm using a general image formation model
(1995) Nicolls, Frederick C; De Jager, Gerhard; Sewell, Bryan Trevor
This were outlines the development of a general imaging model for use in autofocus, astigmatism correction, and resolution analysis. The model is based on the modulation transfer function of the imaging system in the presence of aberrations, in particular defocus. The extension of the model to include astigmatism is also included. The signals used are related to the ratios of the Fourier transforms of images captured under different operating conditions. Methods are developed for working with these signals in a consistent manner. The model described is then applied to the problem of autofocus. A general autofocus algorithm is presented and results given which reﬂect the predictive properties of this model. The imaging system used for the generation of results was a scanning electron microscope, although the conclusions should be valid across a far wider range of instruments. It is however the speciﬁc requirements of the SEM that make the generalisation presented here particularly useful.
Open Access
The Development of a vision-based flotation froth analysis system
(1999) Wright, Benedict Anson; De Jager, Gerhard
This dissertation describes the development of a machine vision system for the on-line analysis of flotation froth images. The size and shape of bubbles that constitute the flotation froth convey considerable information on the performance of the flotation process. A method whereby this size and shape information can be automatically extracted from froth images is highly desirable. In this research, a system was developed which acquires froth image using a video camera, and then rapidly identifies the bubbles in the froth by segmenting the image using a morphological operation known as the Fast Watershed Transform. Bubble size and shape information is extracted from the segmented images and can be correlated with metallurgical and other data from concentrator plants in order to elucidate relationships between froth appearance and plant performance. The machine vision system developed was tested on a platinum concentrator plant, and is able to identify and characterise variations in flotation froth appearance, which occur in response to changes in process inputs. The ability of the system to detect changes in bubble size distribution has been found to be particularly useful in detecting process input variations.
Open Access
The Development of a vision-based flotation froth analysis system
(1999) Wright, Benedict Anson; De Jager, Gerhard
This dissertation describes the development of a machine vision system for the on-line analysis of flotation froth images. The size and shape of bubbles that constitute the flotation froth convey considerable information on the performance of the flotation process. A method whereby this size and shape information can be automatically extracted from froth images is highly desirable. In this research, a system was developed which acquires froth image using a video camera, and then rapidly identifies the bubbles in the froth by segmenting the image using a morphological operation known as the Fast Watershed Transform. Bubble size and shape information is extracted from the segmented images and can be correlated with metallurgical and other data from concentrator plants in order to elucidate relationships between froth appearance and plant performance. The machine vision system developed was tested on a platinum concentrator plant, and is able to identify and characterise variations in flotation froth appearance, which occur in response to changes in process inputs. The ability of the system to detect changes in bubble size distribution has been found to be particularly useful in detecting process input variations.
Open Access
Evaluating the Performance of the Resampled nSight-2 Sensor's Spectral Configuration in Discriminating Wetland Plant Species Using Advanced Classifiers and Spectroscopy.
(2022) Gasela, Mchasisi; De Jager, Gerhard; Kganyago, Mahlatse L
Accurate and reliable information about wetland plant species is critical, as it informs improved preservation, conservation and management of wetland ecosystems. Well managed ecosystems guarantee achieving Sustainable Development Goals. Therefore, remote sensing technique has gained prominence in providing such information. However, broadband sensors are affected by effects of soil and water reflectances associated with wetlands hence cannot adequately discern subtle differences among wetland plant species. On the other hand, hyperspectral sensors allow for an in-depth examination of plant leaf and canopy biochemical traits such as lignin, cellulose, nitrogen, chlorophyll, carotenoids, anthocyanin and water content through spectral measurements which is critical for plant species discrimination. This study sought to test the capability of the forthcoming nSight-2 hyperspectral sensor in discriminating among four dominant wetland plant species. To accomplish this, the performance of nSight-2 spectral settings were compared with those of the upcoming EnMap hyperspectral satellite and an already established Worldview-2 multi-spectral sensor that carries strategic wavebands for vegetation studies, i.e. red-edge and near-infrared. The study also evaluated the performances of non-parametric machine learning algorithms in classifying wetland plant species using nSight-2 spectral configuration. The results showed a high discrimination accuracy by nSight-2 spectral settings with an overall accuracy of 84.09%, followed by Worldview-2 i.e. 81.82% while EnMap was the worst i.e. 77.77%. The most important bands for vegetation analysis were within the visible (VIS), Red-edge (RE) and near infrared (NIR) regions of the electromagnetic spectrum. The study also demonstrated that within these spectral bands, the four dominant Verloren Vallei Nature Reserve wetland plant i.e. Crocosmia sp., Grasses, Agapanthus sp. and Cyperus sp. could be differentiated using the spectral settings of these sensors. Furthermore, the results showed a superior performance of Support Vector Machine (SVM) with overall accuracy of 93.18%, compared with the RF and Partial Least Squares-Discriminant Analysis (PLS-DA) that had overall accuracies of 84.09% and 83.63% respectively. In summary, the study demonstrated that the spectral configuration of nSight-2 hyperspectral sensor can discriminate among the wetland plant species with comparable accuracy to that of a stateof-the-art sensor, i.e. Worldview-2 and better than the upcoming EnMap.
Open Access
An examination of block motion compensation algorithms for MPEG-2 and prediction of bit rates from video sequence measurements
(1997) Francis, Jerome Jonathan; De Jager, Gerhard
This dissertation examines the following two problems: • Finding a block motion compensation algorithm which is optimum in performance and speed. • Predicting the performance, for complex sequences, of an MPEG-2 encoder. An optimum motion compensation algorithm can lead to optimum temporal compression. For fixed bit-rate encoders finding methods to predict the bit-rate from properties of the video sequence can lead to an optimum use of the transmission bandwidth. The examination of motion compensation algorithms involved examining previous algorithms. Historically, one of three functions are used to evaluate a candidate motion vector, namely, Mean Square Error (MSE), Minimum Absolute Difference (MAD) and cross correlation. The ideal motion vector being the one that minimises MAD and MSE, and maximises cross-correlation. Sub-sampling, hierarchical and feature domain methods were examined. Finally some new algorithms are proposed and further areas of research suggested. The new algorithms suggested perform close to optimum, particularly those algorithms searching feature space.
Open Access
Fast implementation of integer transforms in H.264/AVC video encoders
(2011) Lubobya, Smart Charles; Dlodlo, Mqhele E; De Jager, Gerhard
The Integer Discrete Cosine Transform (IDCT) and Hadamard transforms are adopted in the H.264/AVC standard encoder for the compression of residual video signals. Other video standards such as the H.261, H.262 and H.263 use Discrete Cosine Transform (DCT).
Open Access
Fast implementations of block motion estimation algorithms in video encoders
(2011) Koduri, Naga Rohini; Dlodlo, Mqhele E; De Jager, Gerhard
This research is aimed at designing and implementing novel fast algorithms for speeding up the encoding process. The objective of this study was to come up with algorithms which can estimate the data significantly faster than the existing algorithms, whilst ensuring acceptable video quality.
Open Access
Fixed pattern noise compensation in a mercury cadmium telluride infrared focal plane array
(1998) Reddy, Praven; De Jager, Gerhard
This thesis describes techniques for the correction of spatial noise artifacts in a mercury cadmium telluride infrared camera system. The spatial noise artifacts are a result of nonuniformities within the infrared focal plane detector array. The techniques presented dispense with the need for traditional temperature references, and provide nonuniformity compensation by using only the statistics of the moving infrared scene and motion of the camera assembly for calibration. Frame averaging is employed, assuming that all of the detector pixels will eventually be irradiated with the same levels of incident flux after some extended period of time. Using a statistical analysis of the camera image data, the correction coefficients are re-calculated and updated. These techniques also ensure that the calculated coefficients continually track the variations in the dark currents as well as temperature changes within the dewar sensor cooling vessel. These scene-based reference free approaches to the calculation of compensation coefficients in the infrared camera are shown to be successful in compensating for the effects of fixed pattern spatial noise.
Open Access
Fractal image compression and the self-affinity assumption : a stochastic signal modelling perspective
(1996) Wohlberg, Brendt; De Jager, Gerhard
Fractal image compression is a comparatively new technique which has gained considerable attention in the popular technical press, and more recently in the research literature. The most significant advantages claimed are high reconstruction quality at low coding rates, rapid decoding, and "resolution independence" in the sense that an encoded image may be decoded at a higher resolution than the original. While many of the claims published in the popular technical press are clearly extravagant, it appears from the rapidly growing body of published research that fractal image compression is capable of performance comparable with that of other techniques enjoying the benefit of a considerably more robust theoretical foundation. . So called because of the similarities between the form of image representation and a mechanism widely used in generating deterministic fractal images, fractal compression represents an image by the parameters of a set of affine transforms on image blocks under which the image is approximately invariant. Although the conditions imposed on these transforms may be shown to be sufficient to guarantee that an approximation of the original image can be reconstructed, there is no obvious theoretical reason to expect this to represent an efficient representation for image coding purposes. The usual analogy with vector quantisation, in which each image is considered to be represented in terms of code vectors extracted from the image itself is instructive, but transforms the fundamental problem into one of understanding why this construction results in an efficient codebook. The signal property required for such a codebook to be effective, termed "self-affinity", is poorly understood. A stochastic signal model based examination of this property is the primary contribution of this dissertation. The most significant findings (subject to some important restrictions} are that "self-affinity" is not a natural consequence of common statistical assumptions but requires particular conditions which are inadequately characterised by second order statistics, and that "natural" images are only marginally "self-affine", to the extent that fractal image compression is effective, but not more so than comparable standard vector quantisation techniques.