Automatically Evolving Rotation-invariant Texture Image Descriptors by Genetic Programming

In computer vision, training a model that performs classification effectively is highly dependent on the extracted features, and the number of training instances. Conventionally, feature detection and extraction are performed by a domain-expert who, in many cases, is expensive to employ and hard to find. Therefore, image descriptors have emerged to automate these tasks. However, designing an image descriptor still requires domain-expert intervention. Moreover, the majority of machine learning algorithms require a large number of training examples to perform well. However, labelled data is not always available or easy to acquire, and dealing with a large dataset can dramatically slow down the training process. In this paper, we propose a novel Genetic Programming based method that automatically synthesises a descriptor using only two training instances per class. The proposed method combines arithmetic operators to evolve a model that takes an image and generates a feature vector. The performance of the proposed method is assessed using six datasets for texture classification with different degrees of rotation, and is compared with seven domain-expert designed descriptors. The results show that the proposed method is robust to rotation, and has significantly outperformed, or achieved a comparable performance to, the baseline methods.

Impact of Curling and Warping on Concrete Pavement, TR-668, 2016

Portland cement concrete (PCC) pavement undergoes repeated environmental load-related deflection resulting from temperature and moisture variations across the pavement depth. This phenomenon, referred to as PCC pavement curling and warping, has been known and studied since the mid-1920s. Slab curvature can be further magnified under repeated traffic loads and may ultimately lead to fatigue failures, including top-down and bottom-up transverse, longitudinal, and corner cracking. It is therefore important to measure the “true” degree of curling and warping in PCC pavements, not only for quality control (QC) and quality assurance (QA) purposes, but also to achieve a better understanding of its relationship to long-term pavement performance. In order to better understand the curling and warping behavior of PCC pavements in Iowa and provide recommendations to mitigate curling and warping deflections, field investigations were performed at six existing sites during the late fall of 2015. These sites included PCC pavements with various ages, slab shapes, mix design aspects, and environmental conditions during construction. A stationary light detection and ranging (LiDAR) device was used to scan the slab surfaces. The degree of curling and warping along the longitudinal, transverse, and diagonal directions was calculated for the selected slabs based on the point clouds acquired using LiDAR. The results and findings are correlated to variations in pavement performance, mix design, pavement design, and construction details at each site. Recommendations regarding how to minimize curling and warping are provided based on a literature review and this field study. Some examples of using point cloud data to build three-dimensional (3D) models of the overall curvature of the slab shape are presented to show the feasibility of using this 3D analysis method for curling and warping analysis.

A method to determine the impact of reduced visual function on nodule detection performance

Purpose - In this study we aim to validate a method to assess the impact of reduced visual function and observer performance concurrently with a nodule detection task. Materials and methods - Three consultant radiologists completed a nodule detection task under three conditions: without visual defocus (0.00 Dioptres; D), and with two different magnitudes of visual defocus (−1.00 D and −2.00 D). Defocus was applied with lenses and visual function was assessed prior to each image evaluation. Observers evaluated the same cases on each occasion; this comprised of 50 abnormal cases containing 1–4 simulated nodules (5, 8, 10 and 12 mm spherical diameter, 100 HU) placed within a phantom, and 25 normal cases (images containing no nodules). Data was collected under the free-response paradigm and analysed using Rjafroc. A difference in nodule detection performance would be considered significant at p < 0.05. Results - All observers had acceptable visual function prior to beginning the nodule detection task. Visual acuity was reduced to an unacceptable level for two observers when defocussed to −1.00 D and for one observer when defocussed to −2.00 D. Stereoacuity was unacceptable for one observer when defocussed to −2.00 D. Despite unsatisfactory visual function in the presence of defocus we were unable to find a statistically significant difference in nodule detection performance (F(2,4) = 3.55, p = 0.130). Conclusion - A method to assess visual function and observer performance is proposed. In this pilot evaluation we were unable to detect any difference in nodule detection performance when using lenses to reduce visual function.

Comparison of 2.3 & 5 mega pixel (MP) resolution monitors when detecting mammography image blurring

Background - Image blurring in Full Field Digital Mammography (FFDM) is reported to be a problem within many UK breast screening units resulting in significant proportion of technical repeats/recalls. Our study investigates monitors of differing pixel resolution, and whether there is a difference in blurring detection between a 2.3 MP technical review monitor and a 5MP standard reporting monitor. Methods - Simulation software was created to induce different magnitudes of blur on 20 artifact free FFDM screening images. 120 blurred and non-blurred images were randomized and displayed on the 2.3 and 5MP monitors; they were reviewed by 28 trained observers. Monitors were calibrated to the DICOM Grayscale Standard Display Function. T-test was used to determine whether significant differences exist in blurring detection between the monitors. Results - The blurring detection rate on the 2.3MP monitor for 0.2, 0.4, 0.6, 0.8 and 1 mm blur was 46, 59, 66, 77and 78% respectively; and on the 5MP monitor 44, 70, 83 , 96 and 98%. All the non-motion images were identified correctly. A statistical difference (p <0.01) in the blurring detection rate between the two monitors was demonstrated. Conclusions - Given the results of this study and knowing that monitors as low as 1 MP are used in clinical practice, we speculate that technical recall/repeat rates because of blurring could be reduced if higher resolution monitors are used for technical review at the time of imaging. Further work is needed to determine monitor minimum specification for visual blurring detection.

Anomaly Detection in Noisy Images

Finding rare events in multidimensional data is an important detection problem that has applications in many fields, such as risk estimation in insurance industry, finance, flood prediction, medical diagnosis, quality assurance, security, or safety in transportation. The occurrence of such anomalies is so infrequent that there is usually not enough training data to learn an accurate statistical model of the anomaly class. In some cases, such events may have never been observed, so the only information that is available is a set of normal samples and an assumed pairwise similarity function. Such metric may only be known up to a certain number of unspecified parameters, which would either need to be learned from training data, or fixed by a domain expert. Sometimes, the anomalous condition may be formulated algebraically, such as a measure exceeding a predefined threshold, but nuisance variables may complicate the estimation of such a measure. Change detection methods used in time series analysis are not easily extendable to the multidimensional case, where discontinuities are not localized to a single point. On the other hand, in higher dimensions, data exhibits more complex interdependencies, and there is redundancy that could be exploited to adaptively model the normal data. In the first part of this dissertation, we review the theoretical framework for anomaly detection in images and previous anomaly detection work done in the context of crack detection and detection of anomalous components in railway tracks. In the second part, we propose new anomaly detection algorithms. The fact that curvilinear discontinuities in images are sparse with respect to the frame of shearlets, allows us to pose this anomaly detection problem as basis pursuit optimization. Therefore, we pose the problem of detecting curvilinear anomalies in noisy textured images as a blind source separation problem under sparsity constraints, and propose an iterative shrinkage algorithm to solve it. Taking advantage of the parallel nature of this algorithm, we describe how this method can be accelerated using graphical processing units (GPU). Then, we propose a new method for finding defective components on railway tracks using cameras mounted on a train. We describe how to extract features and use a combination of classifiers to solve this problem. Then, we scale anomaly detection to bigger datasets with complex interdependencies. We show that the anomaly detection problem naturally fits in the multitask learning framework. The first task consists of learning a compact representation of the good samples, while the second task consists of learning the anomaly detector. Using deep convolutional neural networks, we show that it is possible to train a deep model with a limited number of anomalous examples. In sequential detection problems, the presence of time-variant nuisance parameters affect the detection performance. In the last part of this dissertation, we present a method for adaptively estimating the threshold of sequential detectors using Extreme Value Theory on a Bayesian framework. Finally, conclusions on the results obtained are provided, followed by a discussion of possible future work.

Lesion detection performance: comparative analysis of low-dose CT data of the chest on two hybrid imaging systems

Incidental findings on low-dose CT images obtained during hybrid imaging are an increasing phenomenon as CT technology advances. Understanding the diagnostic value of incidental findings along with the technical limitations is important when reporting image results and recommending follow-up, which may result in an additional radiation dose from further diagnostic imaging and an increase in patient anxiety. This study assessed lesions incidentally detected on CT images acquired for attenuation correction on two SPECT/CT systems. Methods: An anthropomorphic chest phantom containing simulated lesions of varying size and density was imaged on an Infinia Hawkeye 4 and a Symbia T6 using the low-dose CT settings applied for attenuation correction acquisitions in myocardial perfusion imaging. Twenty-two interpreters assessed 46 images from each SPECT/CT system (15 normal images and 31 abnormal images; 41 lesions). Data were evaluated using a jackknife alternative free-response receiver-operating-characteristic analysis (JAFROC). Results: JAFROC analysis showed a significant difference (P < 0.0001) in lesion detection, with the figures of merit being 0.599 (95% confidence interval, 0.568, 0.631) and 0.810 (95% confidence interval, 0.781, 0.839) for the Infinia Hawkeye 4 and Symbia T6, respectively. Lesion detection on the Infinia Hawkeye 4 was generally limited to larger, higher-density lesions. The Symbia T6 allowed improved detection rates for midsized lesions and some lower-density lesions. However, interpreters struggled to detect small (5 mm) lesions on both image sets, irrespective of density. Conclusion: Lesion detection is more reliable on low-dose CT images from the Symbia T6 than from the Infinia Hawkeye 4. This phantom-based study gives an indication of potential lesion detection in the clinical context as shown by two commonly used SPECT/CT systems, which may assist the clinician in determining whether further diagnostic imaging is justified.

Development of a real time Image analysis system for traffic monitoring applications.

Image processing offers unparalleled potential for traffic monitoring and control. For many years engineers have attempted to perfect the art of automatic data abstraction from sequences of video images. This paper outlines a research project undertaken at Napier University by the authors in the field of image processing for automatic traffic analysis. A software based system implementing TRIP algorithms to count cars and measure vehicle speed has been developed by members of the Transport Engineering Research Unit (TERU) at the University. The TRIP algorithm has been ported and evaluated on an IBM PC platform with a view to hardware implementation of the pre-processing routines required for vehicle detection. Results show that a software based traffic counting system is realisable for single window processing. Due to the high volume of data required to be processed for full frames or multiple lanes, system operations in real time are limited. Therefore specific hardware is required to be designed. The paper outlines a hardware design for implementation of inter-frame and background differencing, background updating and shadow removal techniques. Preliminary results showing the processing time and counting accuracy for the routines implemented in software are presented and a real time hardware pre-processing architecture is described.

Plane Detection Using Affine Homography

AIRES, Kelson R. T.; ARAÚJO, Hélder J.; MEDEIROS, Adelardo A. D. Plane Detection Using Affine Homography. In: CONGRESSO BRASILEIRO DE AUTOMÁTICA, 2008, Juiz de Fora, MG: Anais... do CBA 2008.

Detection of mesoscale thermal fronts from 4km data using smoothing techniques: Gradient-based fronts classification and basin scale application

In order to optimize frontal detection in sea surface temperature fields at 4 km resolution, a combined statistical and expert-based approach is applied to test different spatial smoothing of the data prior to the detection process. Fronts are usually detected at 1 km resolution using the histogram-based, single image edge detection (SIED) algorithm developed by Cayula and Cornillon in 1992, with a standard preliminary smoothing using a median filter and a 3 × 3 pixel kernel. Here, detections are performed in three study regions (off Morocco, the Mozambique Channel, and north-western Australia) and across the Indian Ocean basin using the combination of multiple windows (CMW) method developed by Nieto, Demarcq and McClatchie in 2012 which improves on the original Cayula and Cornillon algorithm. Detections at 4 km and 1 km of resolution are compared. Fronts are divided in two intensity classes (“weak” and “strong”) according to their thermal gradient. A preliminary smoothing is applied prior to the detection using different convolutions: three type of filters (median, average and Gaussian) combined with four kernel sizes (3 × 3, 5 × 5, 7 × 7, and 9 × 9 pixels) and three detection window sizes (16 × 16, 24 × 24 and 32 × 32 pixels) to test the effect of these smoothing combinations on reducing the background noise of the data and therefore on improving the frontal detection. The performance of the combinations on 4 km data are evaluated using two criteria: detection efficiency and front length. We find that the optimal combination of preliminary smoothing parameters in enhancing detection efficiency and preserving front length includes a median filter, a 16 × 16 pixel window size, and a 5 × 5 pixel kernel for strong fronts and a 7 × 7 pixel kernel for weak fronts. Results show an improvement in detection performance (from largest to smallest window size) of 71% for strong fronts and 120% for weak fronts. Despite the small window used (16 × 16 pixels), the length of the fronts has been preserved relative to that found with 1 km data. This optimal preliminary smoothing and the CMW detection algorithm on 4 km sea surface temperature data are then used to describe the spatial distribution of the monthly frequencies of occurrence for both strong and weak fronts across the Indian Ocean basin. In general strong fronts are observed in coastal areas whereas weak fronts, with some seasonal exceptions, are mainly located in the open ocean. This study shows that adequate noise reduction done by a preliminary smoothing of the data considerably improves the frontal detection efficiency as well as the global quality of the results. Consequently, the use of 4 km data enables frontal detections similar to 1 km data (using a standard median 3 × 3 convolution) in terms of detectability, length and location. This method, using 4 km data is easily applicable to large regions or at the global scale with far less constraints of data manipulation and processing time relative to 1 km data.

Leveraging Multiple Features for Image Retrieval and Matching

The goal of image retrieval and matching is to find and locate object instances in images from a large-scale image database. While visual features are abundant, how to combine them to improve performance by individual features remains a challenging task. In this work, we focus on leveraging multiple features for accurate and efficient image retrieval and matching. We first propose two graph-based approaches to rerank initially retrieved images for generic image retrieval. In the graph, vertices are images while edges are similarities between image pairs. Our first approach employs a mixture Markov model based on a random walk model on multiple graphs to fuse graphs. We introduce a probabilistic model to compute the importance of each feature for graph fusion under a naive Bayesian formulation, which requires statistics of similarities from a manually labeled dataset containing irrelevant images. To reduce human labeling, we further propose a fully unsupervised reranking algorithm based on a submodular objective function that can be efficiently optimized by greedy algorithm. By maximizing an information gain term over the graph, our submodular function favors a subset of database images that are similar to query images and resemble each other. The function also exploits the rank relationships of images from multiple ranked lists obtained by different features. We then study a more well-defined application, person re-identification, where the database contains labeled images of human bodies captured by multiple cameras. Re-identifications from multiple cameras are regarded as related tasks to exploit shared information. We apply a novel multi-task learning algorithm using both low level features and attributes. A low rank attribute embedding is joint learned within the multi-task learning formulation to embed original binary attributes to a continuous attribute space, where incorrect and incomplete attributes are rectified and recovered. To locate objects in images, we design an object detector based on object proposals and deep convolutional neural networks (CNN) in view of the emergence of deep networks. We improve a Fast RCNN framework and investigate two new strategies to detect objects accurately and efficiently: scale-dependent pooling (SDP) and cascaded rejection classifiers (CRC). The SDP improves detection accuracy by exploiting appropriate convolutional features depending on the scale of input object proposals. The CRC effectively utilizes convolutional features and greatly eliminates negative proposals in a cascaded manner, while maintaining a high recall for true objects. The two strategies together improve the detection accuracy and reduce the computational cost.

Automatic Glaucoma Detection Based on Optic Disc Segmentation and Texture Feature Extraction

The use of digital image processing techniques is prominent in medical settings for the automatic diagnosis of diseases. Glaucoma is the second leading cause of blindness in the world and it has no cure. Currently, there are treatments to prevent vision loss, but the disease must be detected in the early stages. Thus, the objective of this work is to develop an automatic detection method of Glaucoma in retinal images. The methodology used in the study were: acquisition of image database, Optic Disc segmentation, texture feature extraction in different color models and classification of images in glaucomatous or not. We obtained results of 93% accuracy

Wind turbine rotor blade monitoring using digital image correlation: a comparison to aeroelastic simulations of a multi-megawatt wind turbine

Optical full-field measurement methods such as Digital Image Correlation (DIC) provide a new opportunity for measuring deformations and vibrations with high spatial and temporal resolution. However, application to full-scale wind turbines is not trivial. Elaborate preparation of the experiment is vital and sophisticated post processing of the DIC results essential. In the present study, a rotor blade of a 3.2 MW wind turbine is equipped with a random black-and-white dot pattern at four different radial positions. Two cameras are located in front of the wind turbine and the response of the rotor blade is monitored using DIC for different turbine operations. In addition, a Light Detection and Ranging (LiDAR) system is used in order to measure the wind conditions. Wind fields are created based on the LiDAR measurements and used to perform aeroelastic simulations of the wind turbine by means of advanced multibody codes. The results from the optical DIC system appear plausible when checked against common and expected results. In addition, the comparison of relative out-of-plane blade deflections shows good agreement between DIC results and aeroelastic simulations.

Hand detection on images based on deformable part models and additional features

Hand detection on images has important applications on person activities recognition. This thesis focuses on PASCAL Visual Object Classes (VOC) system for hand detection. VOC has become a popular system for object detection, based on twenty common objects, and has been released with a successful deformable parts model in VOC2007. A hand detection on an image is made when the system gets a bounding box which overlaps with at least 50% of any ground truth bounding box for a hand on the image. The initial average precision of this detector is around 0.215 compared with a state-of-art of 0.104; however, color and frequency features for detected bounding boxes contain important information for re-scoring, and the average precision can be improved to 0.218 with these features. Results show that these features help on getting higher precision for low recall, even though the average precision is similar.

The role of context in image annotation and recommendation

With the rise of smart phones, lifelogging devices (e.g. Google Glass) and popularity of image sharing websites (e.g. Flickr), users are capturing and sharing every aspect of their life online producing a wealth of visual content. Of these uploaded images, the majority are poorly annotated or exist in complete semantic isolation making the process of building retrieval systems difficult as one must firstly understand the meaning of an image in order to retrieve it. To alleviate this problem, many image sharing websites offer manual annotation tools which allow the user to “tag” their photos, however, these techniques are laborious and as a result have been poorly adopted; Sigurbjörnsson and van Zwol (2008) showed that 64% of images uploaded to Flickr are annotated with < 4 tags. Due to this, an entire body of research has focused on the automatic annotation of images (Hanbury, 2008; Smeulders et al., 2000; Zhang et al., 2012a) where one attempts to bridge the semantic gap between an image’s appearance and meaning e.g. the objects present. Despite two decades of research the semantic gap still largely exists and as a result automatic annotation models often offer unsatisfactory performance for industrial implementation. Further, these techniques can only annotate what they see, thus ignoring the “bigger picture” surrounding an image (e.g. its location, the event, the people present etc). Much work has therefore focused on building photo tag recommendation (PTR) methods which aid the user in the annotation process by suggesting tags related to those already present. These works have mainly focused on computing relationships between tags based on historical images e.g. that NY and timessquare co-exist in many images and are therefore highly correlated. However, tags are inherently noisy, sparse and ill-defined often resulting in poor PTR accuracy e.g. does NY refer to New York or New Year? This thesis proposes the exploitation of an image’s context which, unlike textual evidences, is always present, in order to alleviate this ambiguity in the tag recommendation process. Specifically we exploit the “what, who, where, when and how” of the image capture process in order to complement textual evidences in various photo tag recommendation and retrieval scenarios. In part II, we combine text, content-based (e.g. # of faces present) and contextual (e.g. day-of-the-week taken) signals for tag recommendation purposes, achieving up to a 75% improvement to precision@5 in comparison to a text-only TF-IDF baseline. We then consider external knowledge sources (i.e. Wikipedia & Twitter) as an alternative to (slower moving) Flickr in order to build recommendation models on, showing that similar accuracy could be achieved on these faster moving, yet entirely textual, datasets. In part II, we also highlight the merits of diversifying tag recommendation lists before discussing at length various problems with existing automatic image annotation and photo tag recommendation evaluation collections. In part III, we propose three new image retrieval scenarios, namely “visual event summarisation”, “image popularity prediction” and “lifelog summarisation”. In the first scenario, we attempt to produce a rank of relevant and diverse images for various news events by (i) removing irrelevant images such memes and visual duplicates (ii) before semantically clustering images based on the tweets in which they were originally posted. Using this approach, we were able to achieve over 50% precision for images in the top 5 ranks. In the second retrieval scenario, we show that by combining contextual and content-based features from images, we are able to predict if it will become “popular” (or not) with 74% accuracy, using an SVM classifier. Finally, in chapter 9 we employ blur detection and perceptual-hash clustering in order to remove noisy images from lifelogs, before combining visual and geo-temporal signals in order to capture a user’s “key moments” within their day. We believe that the results of this thesis show an important step towards building effective image retrieval models when there lacks sufficient textual content (i.e. a cold start).