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.

MICROFLUIDIC ASSAY PERFORMANCE ENHANCEMENT USING POROUS VOLUMETRIC DETECTION ELEMENTS FOR IMPEDEMETRIC AND OPTICAL SENSING

Microfluidic technologies have great potential to help create automated, cost-effective, portable devices for rapid point of care (POC) diagnostics in diverse patient settings. Unfortunately commercialization is currently constrained by the materials, reagents, and instrumentation required and detection element performance. While most microfluidic studies utilize planar detection elements, this dissertation demonstrates the utility of porous volumetric detection elements to improve detection sensitivity and reduce assay times. Impedemetric immunoassays were performed utilizing silver enhanced gold nanoparticle immunoconjugates (AuIgGs) and porous polymer monolith or silica bead bed detection elements within a thermoplastic microchannel. For a direct assay with 10 µm spaced electrodes the detection limit was 0.13 fM AuIgG with a 3 log dynamic range. The same assay was performed with electrode spacing of 15, 40, and 100 µm with no significant difference between configurations. For a sandwich assay the detection limit was10 ng/mL with a 4 log dynamic range. While most impedemetric assays rely on expensive high resolution electrodes to enhance planar senor performance, this study demonstrates the employment of porous volumetric detection elements to achieve similar performance using lower resolution electrodes and shorter incubation times. Optical immunoassays were performed using porous volumetric capture elements perfused with refractive index matching solutions to limit light scattering and enhance signal. First, fluorescence signal enhancement was demonstrated with a porous polymer monolith within a silica capillary. Next, transmission enhancement of a direct assay was demonstrated by infusing aqueous sucrose solutions through silica bead beds with captured silver enhanced AuIgGs yielding a detection limit of 0.1 ng/mL and a 5 log dynamic range. Finally, ex situ functionalized porous silica monolith segments were integrated into thermoplastic channels for a reflectance based sandwich assay yielding a detection limit of 1 ng/mL and a 5 log dynamic range. The simple techniques for optical signal enhancement and ex situ element integration enable development of sensitive, multiplexed microfluidic sensors. Collectively the demonstrated experiments validate the use of porous volumetric detection elements to enhance impedemetric and optical microfluidic assays. The techniques rely on commercial reagents, materials compatible with manufacturing, and measurement instrumentation adaptable to POC diagnostics.