Improve Retrieval Precision Research Articles

An Ontological Chinese Legal Consultation System

We propose an ontological Chinese legal consultation system (CLCS) based on Chinese legal characteristics by integrating statutes and judicial precedents to facilitate the retrieval of relevant statutes and judgments for the general public. Ontology, an emerging research topic in recent years, incorporates a hierarchical structure and supports logical reasoning, which can reduce semantic ambiguities and extract implied semantic information. We constructed statutes ontology and a case ontology using a bottom-up method and a top-down method, respectively. Then, to test the retrieval precision of the proposed CLCS, we performed experiments using emotional damage compensation as an example by combining genetic algorithm and k-nearest neighbor approaches. Experimental results demonstrate that the proposed method improves retrieval precision.

A 3D shape matching and retrieval approach based on fusion of curvature and geometric diffusion features

The majority of shape matching and retrieval methods use only one single shape descriptor. Unfortunately, no shape descriptor is sufficient to provide suitable results for all kinds of shapes. The most common way to improve the performance of shape descriptors is to fuse them. In this paper, we propose a new 3D matching and retrieval approach based on a fully unsupervised fusion of curvature and geometric diffusion descriptors. In fact, to improve retrieval precision, we use two descriptors based on local and global features extracted from a shape, and automatically combine these features using a fusion method called Product rule. The Product rule combines values assigned to vertices by the two descriptors. This fusion rule gives better results compared to other well-known fusion schemes such as Max, Min and Linear rules. The proposed approach improves considerably the retrieval precision even with pose changes. This is shown through the retrieval results obtained on several popular 3D shape benchmarks.

Adaptive fusion of color and spatial features for noise-robust retrieval of colored logo and trademark images

Due to their uniqueness and high value commercially, logos/trademarks play a key role in e-business based global marketing. However, existing trademark/logo retrieval techniques and content-based image retrieval methods are mostly designed for generic images, which cannot provide effective retrieval of trademarks/logos. Although color and spatial features have been intensively investigated for logo image retrieval, in most cases they were applied separately. When these are combined in a fused manner, a fixed weighting is normally used between them which cannot reflect the significance of these features in the images. When the image quality is degraded by various reasons such as noise, the reliability of color and spatial features may change in different ways, such that the weights between them should be adapted to such changes. In this paper, adaptive fusion of color and spatial descriptors is proposed for colored logo/trademark image retrieval. First, color quantization and k-means are combined for effective dominant color extraction. For each extracted dominant color, a component-based spatial descriptor is derived for local features. By analyzing the image histogram, an adaptive fusion of these two features is achieved for more effective logo abstraction and more accurate image retrieval. The proposed approach has been tested on a database containing over 2300 logo/trademark images. Experimental results have shown that the proposed methodology yields improved retrieval precision and outperforms three state-of-the-art techniques even with added Gaussian, salt and pepper, and speckle noise.

Preparing a collection of radiology examinations for distribution and retrieval.

Clinical documents made available for secondary use play an increasingly important role in discovery of clinical knowledge, development of research methods, and education. An important step in facilitating secondary use of clinical document collections is easy access to descriptions and samples that represent the content of the collections. This paper presents an approach to developing a collection of radiology examinations, including both the images and radiologist narrative reports, and making them publicly available in a searchable database. The authors collected 3996 radiology reports from the Indiana Network for Patient Care and 8121 associated images from the hospitals' picture archiving systems. The images and reports were de-identified automatically and then the automatic de-identification was manually verified. The authors coded the key findings of the reports and empirically assessed the benefits of manual coding on retrieval. The automatic de-identification of the narrative was aggressive and achieved 100% precision at the cost of rendering a few findings uninterpretable. Automatic de-identification of images was not quite as perfect. Images for two of 3996 patients (0.05%) showed protected health information. Manual encoding of findings improved retrieval precision. Stringent de-identification methods can remove all identifiers from text radiology reports. DICOM de-identification of images does not remove all identifying information and needs special attention to images scanned from film. Adding manual coding to the radiologist narrative reports significantly improved relevancy of the retrieved clinical documents. The de-identified Indiana chest X-ray collection is available for searching and downloading from the National Library of Medicine (http://openi.nlm.nih.gov/).

Individual strategies in the tasks of graphical retrieval of technical drawings

A designer׳s information need can have a lot of finer points not expressible completely by textual attributes or global aggregated image features. The purpose of the study is to develop a cognitive environment for content-based engineering drawing retrieval, allowing a user to concretize queries, to implement strategies that are most effective for the current search task. We propose meaningful customizable graphic search patterns, various comparison modes and visualization of matched elements. The presented experiments show that the approach yields good results in the sense of relevance and is comfortable in terms of controllability. The possibility to customize search patterns can result in improving retrieval precision by 10–15%. In practice, it can significantly increase the degree of automation of the design-reuse process.

Image Retrieval Method Based on IPDSH and SRIP

At present, the Content-Based Image Retrieval (CBIR) system has become a hot research topic in the computer vision field. In the CBIR system, the accurate extractions of low-level features can reduce the gaps between high-level semantics and improve retrieval precision. This paper puts forward a new retrieval method aiming at the problems of high computational complexities and low precision of global feature extraction algorithms. The establishment of the new retrieval method is on the basis of the SIFT and Harris (APISH) algorithm, and the salient region of interest points (SRIP) algorithm to satisfy users interests in the specific targets of images. In the first place, by using the IPDSH and SRIP algorithms, we tested stable interest points and found salient regions. The interest points in the salient region were named as salient interest points. Secondary, we extracted the pseudo-Zernike moments of the salient interest points’ neighborhood as the feature vectors. Finally, we calculated the similarities between query and database images. Finally, We conducted this experiment based on the Caltech-101 database. By studying the experiment, the results have shown that this new retrieval method can decrease the interference of unstable interest points in the regions of non-interests and improve the ratios of accuracy and recall.

Remote Sensing Image Retrieval with Combined Features Of Salient Region

Abstract. Low-level features tend to achieve unsatisfactory retrieval results in remote sensing image retrieval community because of the existence of semantic gap. In order to improve retrieval precision, visual attention model is used to extract salient objects from image according to their saliency. Then color and texture features are extracted from salient objects and regarded as feature vectors for image retrieval. Experimental results demonstrate that our method improves retrieval results and obtains higher precision.

A Survey on Visual Search Reranking

Search reranking is considered as a best and common way to improves retrieval precision. However the problem is not trivial especially when we are considering multiple features or modalities for search in image and video retrieval. This paper suggests a new kind of reranking algorithm, the circular reranking, that supports the mutual exchange of information across multiple modalities for improving search performance, and follows the philosophy of strong performing modality could learn from weaker ones, and weaker modality does benefit by interacting with stronger modalities. Technically, circular reranking conducts multiple cyclic runs of random walks by exchanging the ranking scores among different features. Contrasting the existing techniques, the reranking procedure reassures interaction between various modalities to seek consensus which are useful for reranking. In this paper, we study several circular reranking properties, including how and which order of information propagation should be configured to fully exploit the potential of modalities for reranking.

Visual words dictionaries and fusion techniques for searching people through textual and visual attributes

Using personal traits for searching people is paramount in several application areas and has attracted an ever-growing attention from the scientific community over the past years. Some practical applications in the realm of digital forensics and surveillance include locating a suspect or finding missing people in a public space. In this paper, we aim at assigning describable visual attributes (e.g., white chubby male wearing glasses and with bangs) as labels to images to describe their appearance and performing visual searches without relying on image annotations during testing. For that, we create mid-level image representations for face images based on visual dictionaries linking visual properties in the images to describable attributes. In addition, we take advantage of machine learning techniques for combining different attributes and performing a query. First, we propose three methods for building the visual dictionaries. Method #1 uses a sparse-sampling scheme to obtain low-level features with a clustering algorithm to build the visual dictionaries. Method #2 uses dense-sampling to obtain low-level features and random selection to build the visual dictionaries while Method #3 uses dense-sampling to obtain low-level features followed by a clustering algorithm to build the visual dictionaries. Thereafter, we train 2-class classifiers for the describable visual attributes of interest which assign to each image a decision score used to obtain its ranking. For more complex queries (2+ attributes), we use three state-of-the-art approaches for combining the rankings: (1) product of probabilities, (2) rank aggregation and (3) rank position. To date, we have considered fifteen attribute classifiers and, consequently, their direct counterparts theoretically allowing 215=32,768 different combined queries (the actual number is smaller since some attributes are contradictory or mutually exclusive). Notwithstanding, the method is easily extensible to include new attributes. Experimental results show that Method #3 greatly improves retrieval precision for some attributes in comparison with other methods in the literature. Finally, for combined attributes, product of probabilities, rank aggregation and rank position yield complementary results for rank fusion and the final decision making suggesting interesting possible combinations for further work.

Image Retrieval Based on Contour and Relevance Feedback

In this paper an algorithm is proposed to retrieve images based on contour moment invariants of image and relevance feedback. Firstly, the contour of each query image is extracted and its contour moment invariant is computed. Then according to Euclid Distance between the query image and each image in the image database, the most similar images to the query image can be found. Finally, the relevance feedback algorithm based on support vector machine (SVM) is applied to improve retrieval precision. Experimental results show that the algorithm is more accurate and efficient to retrieve images with monotonous background and clear object and meet the invariance on shift, rotation and scale transform of objects.

Forest Canopy Cover and Height From MISR in Topographically Complex Southwestern US Landscapes Assessed With High Quality Reference Data

This study addresses the retrieval of spatially contiguous canopy cover and height estimates in southwestern US forests via inversion of a geometric-optical (GO) model against surface bidirectional reflectance factor (BRF) estimates from the Multi-angle Imaging SpectroRadiometer (MISR). Model inversion can provide such maps if good estimates of the background bidirectional reflectance distribution function (BRDF) are avail- able. The study area is in the Sierra National Forest in the Sierra Nevada of California. Tree number density, mean crown radius, and fractional cover reference estimates were obtained via analysis of QuickBird 0.6 m spatial resolution panchromatic imagery using the CANopy Analysis with Panchromatic Imagery (CANAPI) algorithm, while RH50, RH75 and RH100 (50%, 75%, and 100% energy return) height data were obtained from the NASA Laser Vegetation Imaging Sensor (LVIS), a full waveform light detection and ranging (lidar) instrument. These canopy parameters were used to drive a modified version of the simple GO model (SGM), accurately reproducing patterns of MISR 672 nm band surface reflectance (mean RMSE = 0.011, mean R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> = 0.82, N = 1048). Cover and height maps were obtained through model inversion against MISR 672 nm reflectance estimates on a 250 m grid. The free parameters were tree number density and mean crown radius. RMSE values with respect to reference data for the cover and height retrievals were 0.05 and 6.65 m, respectively, with R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.54 and 0.49. MISR can thus provide maps of forest cover and height in areas of topographic variation although refinements are required to improve retrieval precision.

烟草叶面积指数的高光谱估算模型

Leaf area index(LAI) is an important biophysical parameter and is a critical variable in many ecology models,productivity models,and carbon circulation studies.In the present study,we aimed to assess and compare various hyperspectral models in terms of their prediction power of tobacco LAI.In a pot experiment,tobacco canopy hyperspectral reflectance data of the root extending stage,fast growing stage,and mature stage under different water and nitrogen levels were collected with an ASD FieldSpec HandHeld spectroradiometer,Corresponding tobacco LAI values were also collected.LAI retrieval methods were evaluated using four vegetation indices: Ratio vegetation Index(RVI),Normalized difference vegetation index(NDVI),Modified soil-adjusted vegetation index(MSAVI) and Modified second triangular vegetation index(MTVI2),and using principal component analysis(PCA),and neural network(NN) methods.The LAI estimations of these methods were then compared.Results indicated that these methods can make robust estimates of LAI.Determination coefficients(R2) of the validated models of the vegetation indices,PCA,and NN were 0.768-0.852,0.938 and 0.889,respectively.The PCA and NN methods showed higher precision.The stability of the PCA validated model is the best because its root mean square error(RMSE) of 0.172 is smaller than those of the vegetation indices(0.237-0.322) and NN(0.195).Among the vegetation indices,MTVI2 and MSAVI could remove the influence of soil and atmosphere and obtain better retrieval accuracy than either RVI or NDVI.Overall,the PCA and NN methods could improve retrieval precision and therefore be selected for LAI estimation. The vegetation indices achieved a good level of accuracy in estimating tobacco LAI;however,as they generally use information from only a few wavelengths,model stability cannot be guaranteed.However,the regression model based on the vegetation indices does not require a large sample to be assured of accuracy within a certain range.The PCA method can effectively reduce the number of dimensions and retain the important information.The two main components transformed by PCA can be interpreted as the visible spectral factor and near-infrared spectral factor,which represented 95.71% of the variation in the hyper spectral data.PCA can make use of the complementary advantages of different spectral bands,and lower the random disturbance caused by some bands.This means that PCA can be a reliable and general method for LAI estimation.The neural network method has a strong nonlinear mapping ability,and does not require normally distributed data.The model can be effectively trained and tested when a large amount of data is available.Although the neural network model has a strong ability for linear and nonlinear fitting,it is unable to provide any insight on the power of potential explanatory variables to explain variation in the data.It is difficult to fully explain the decisions and processes of producing hyper spectral data output.At present,there are no specific rules that can be followed with the band combinations employed.Further study is required to understand the effects of integrating hyperspectral data of the bands after 1050 nm wavelength(for example,from 2500 nm) on the estimation of tobacco LAI with PCA or NN.

Improving 3D similarity search by enhancing and combining 3D descriptors

Effective content-based retrieval in 3D model databases is an important problem that has attracted much research attention over the last years. Many individual methods proposed to date rely on calculating global 3D model descriptors based on image, surface, volumetric, or structural model properties. Descriptors such as these are then input for determining the degree of similarity between models. Traditionally, the ability of individual descriptors to perform effective 3D search is decided by benchmarking. However, in practice the data set on which 3D retrieval is to be applied may differ from the characteristics of the respective benchmark. Therefore, statically determining the descriptor to use based on a fixed benchmark may lead to suboptimal results. We propose a generic strategy to improve the retrieval effectiveness in 3D retrieval systems consisting of multiple model descriptors. The specific contribution of this paper is two-fold. First, we propose to adaptively combine multiple descriptors by forming weighted descriptor combinations, where the weight of each descriptor is decided at query time. Second, we enhance the set of global model descriptors to be combined by including partial descriptors of the same kind in the combinations. Partial descriptors are obtained by applying a given descriptor extractor on the set of parts of a model, obtained by a simple model partitioning scheme. Thereby, more model information is exposed to the 3D descriptors, leading to a more complete object description. We give a systematic discussion of the descriptor combination space involving static and query-adaptive weighting schemes, and based on descriptors of different type and focus (model global vs. partial). The combination of both global and partial model descriptors is shown to deliver improved retrieval precision, compared to policies using single descriptors or fixed-weight combinations. The resulting scheme is generic and can accommodate a large class of global 3D model descriptors.

Improving retrieval precision of NO2 density in DOAS by accurately ascertaining low-frequency structure

The NO2 density error retrieved by the traditional method of differential optical absorption spectroscopy (DOAS) was analyzed. The technique for reducing the error is proposed and used to obtain the NO2 density of Chengdu through DOAS.

Knowledge-Driven Multidimensional Indexing Structure for Biomedical Media Database Retrieval

Today, biomedical media data are being generated at rates unimaginable only years ago. Content-based retrieval of biomedical media from large databases is becoming increasingly important to clinical, research, and educational communities. In this paper, we present the recently developed entropy balanced statistical (EBS) k-d tree and its applications to biomedical media, including a high-resolution computed tomography (HRCT) lung image database and the first real-time protein tertiary structure search engine. Our index utilizes statistical properties inherent in large-scale biomedical media databases for efficient and accurate searches. By applying concepts from pattern recognition and information theory, the EBS k-d tree is built through top-down decision tree induction. Experimentation shows similarity searches against a protein structure database of 53 363 structures consistently execute in less than 8.14 ms for the top 100 most similar structures. Additionally, we have shown improved retrieval precision over adaptive and statistical k-d trees. Retrieval precision of the EBS k-d tree is 81.6% for content-based retrieval of HRCT lung images and 94.9% at 10% recall for protein structure similarity search. The EBS k-d tree has enormous potential for use in biomedical applications embedded with ground-truth knowledge and multidimensional signatures.

A practical ontology query expansion algorithm for semantic-aware learning objects retrieval

Following the rapid development of Internet, particularly web page interaction technology, distant e-learning has become increasingly realistic and popular. To solve the problems associated with sharing and reusing teaching materials in different e-learning systems, several standard formats, including SCORM, IMS, LOM, and AICC, etc., recently have been proposed by several different international organizations. SCORM LOM, namely learning object metadata, facilitates the indexing and searching of learning objects in a learning object repository through extended sharing and searching features. However, LOM suffers a weakness in terms of semantic-awareness capability. Most information retrieval systems assume that users have cognitive ability regarding their needs. However, in e-learning systems, users may have no idea of what they are looking for and the learning object metadata. This study presents an ontological approach for semantic-aware learning object retrieval. This approach has two significant novel features: a fully automatic ontology-based query expansion algorithm for inferring and aggregating user intention based on their original short query, and another “ambiguity removal” procedure for correcting inappropriate user query terms. This approach is sufficiently generic to be embedded to other LOM-based search mechanisms for semantic-aware learning object retrieval. Focused on digital learning material and contrasted to other traditional keyword-based search technologies, the proposed approach has experimentally demonstrated significantly improved retrieval precision and recall rate.

Incorporating Multiple Index Vectors and Feedback to Improve Urban Objects Retrieval

This paper reports results of integrating multiple index vectors and incorporating feedback in retrieving urban objects from high-resolution images. The integrated method combines an index vector derived from a semivariogram method and another index vector from a wavelet transform method. Results show that the integrated method notably improves retrieval precision over either method when used individually. Feedback on whether a retrieved image is relevant is subsequently incorporated into the retrieval. F-statistics is used to evaluate the ability of individual indices to identify relevant images and to weight these indices. The weighted index vector typically improves retrieval performance further by 10% to 15%.