Larry Huston

An efficient parts-based near-duplicate and sub-image retrieval system

We introduce a system for near-duplicate detection and sub-image retrieval. Such a system is useful for finding copyright violations and detecting forged images. We define near-duplicate as images altered with common transformations such as changing contrast, saturation, scaling, cropping, framing, etc. Our system builds a parts-based representation of images using <i>distinctive local descriptors</i> which give high quality matches even under severe transformations. To cope with the large number of features extracted from the images, we employ <i>locality-sensitive hashing</i> to index the local descriptors. This allows us to make approximate similarity queries that only examine a small fraction of the database. Although locality-sensitive hashing has excellent theoretical performance properties, a standard implementation would still be unacceptably slow for this application. We show that, by optimizing layout and access to the index data on disk, we can efficiently query indices containing millions of keypoints. Our system achieves near-perfect accuracy (100% precision at 99.85% recall) on the tests presented in Meng <i>et al.</i> [16], and consistently strong results on our own, significantly more challenging experiments. Query times are interactive even for collections of thousands of images.

Object-based image retrieval using the statistical structure of images

We propose a new Bayesian approach to object-based image retrieval with relevance feedback. Although estimating the object posterior probability density from few examples seems infeasible, we are able to approximate this density by exploiting statistics of the image database domain. Unlike previous approaches that assume an arbitrary distribution for the unconditional density of the feature vector (the density of the features taken over the entire image domain), we learn both the structure and the parameters of this density. These density estimates enable us to construct a Bayesian classifier. Using this Bayesian classifier, we perform a windowed scan over images for objects of interest and employ the users feedback on the search results to train a second classifier that focuses on eliminating difficult false positives. We have incorporated this algorithm into an object-based image retrieval system. We demonstrate the effectiveness of our approach with experiments using a set of categories from the Corel database.

Accelerating database operators using a network processor

Database management systems (DBMSs) do not take full advantage of modern microarchitectural resources, such as wide-issue out-of-order processor pipelines. Increases in processor clock rate and instruction-level parallelism have left memory accesses as the dominant bottleneck in DBMS execution. Prior research indicates that simultaneous multithreading (SMT) can hide memory access latency from a single thread and improve throughput by increasing the number of outstanding memory accesses. Rather than expend chip area and power on out-of-order execution, as in current SMT processors, we demonstrate the effectiveness of using many simple processor cores, each with hardware support for multiple thread contexts. This paper shows an existing hardware architecture—the network processor—already fits the model for multi-threaded, multi-core execution. Using an Intel IXP2400 network processor, we evaluate the performance of three key database operations and demonstrate improvements of 1.9X to 2.5X when compared to a generalpurpose processor.

Forensic video reconstruction

This paper describes an application that enables quick reconstruction of interconnected events, sparsely captured by one or more surveillance cameras. Unlike related efforts, our approach does not require indexing, advance knowledge of potential search criteria, nor a solution to the generalized object-recognition problem. Instead, we strategically pair the intelligence and skill of a human investigator with the speed and flexibility of a parallel image search engine that exploits local storage and processing capabilities distributed across large collections of video recording devices. The result is a system for fast, interactive, brute-force video searching which is both effective and highly scalable.

SnapFind: brute force interactive image retrieval

SnapFind is an image retrieval system that enables efficient interactive search of large data sets by exploiting active disk technology. In contrast to earlier approaches, where data is typically pre-indexed for efficient retrieval according to a fixed scheme, SnapFind provides users with the flexibility to search non-indexed data in a brute force manner. The query is translated into a customized searchlet that is executed in parallel by processors near the storage devices. This enables the majority of irrelevant images to be discarded where they are stored. Partial results are displayed during search execution allowing users to interactively refine the query without waiting for search termination. This paper argues that algorithms with user-adjustable parameters are preferable to black-box image retrieval techniques.

Dynamic load balancing for distributed search

This paper examines how computation can be mapped across the nodes of a distributed search system to effectively utilize available resources. We specifically address computationally intensive search of complex data, such as content-based retrieval of digital images or sounds, where sophisticated algorithms must be evaluated on the objects of interest. Since these problems require significant computation, we distribute the search over a collection of compute nodes, such as active storage devices, intermediate processors and host computers. A key challenge with mapping the desired computation to the available resources is that the most efficient distribution depends on several factors: relative power and number of compute nodes; network bandwidth between the compute nodes; the cost of evaluating query predicates; and the selectivity of the given query. This wide range of variables renders manual partitioning of the computation infeasible, particularly since some of the parameters (e.g., available network bandwidth) can change during the course of a search. This paper proposes several techniques for dynamic partitioning of computation, and demonstrates that they can significantly improve efficiency for distributed search applications.