Efficient Cloud Image Retrieval System Using Weighted-Inverted Index and Database Filtering Algorithms

With the advance of multimedia technology and communications, images and videos become the major streaming information through the Internet. How to fast retrieve desired similar images precisely from the Internet scale image/video databases is the most important retrieval control target. In this paper, a cloud based content-based image retrieval （CBIR） scheme is presented. Database-categorizing based on weighted-inverted index （DCWII） and database f＇dtering algorithm （DFA） is used to speed up the features matching process. In the DCWII, the weights are assigned to discrete cosine transform （DCT） coefficients histograms and the database is categorized by weighted features. In addition, the DFA filters out the irrelevant image in the database to reduce unnecessary computation loading for features matching. Experiments show that the proposed CBIR scheme outperforms previous work in the precision-recall performance and maintains mean average precision （mAP） about 0.678 in the large-scale database comprising one million images. Our scheme also can reduce about 50% to 85% retrieval time by pre-filtering the database, which helps to improve the efficiency of retrieval systems.

Indexing of Content-Based Image Retrieval System with Image Understanding Approach

This paper presents a novel efficient semantic image classification algorithm for high-level feature indexing of high-dimension image database. Experiments show that the algorithm performs well. The size of the train set and the test set is 7 537 and 5 000 respectively. Based on this theory, another ground is built with 12,000 images, which are divided into three classes: city, landscape and person, the total result of the classifications is 88.92%, meanwhile, some preliminary results are presented for image understanding based on semantic image classification and low level features. The groundtruth for the experiments is built with the images from Corel database, photos and some famous face databases.

Deep Root Memory Optimized Indexing Methodology for Image Search Engines

Digitization has created an abundance of new information sources by altering how pictures are captured.Accessing large image databases from a web portal requires an opted indexing structure instead of reducing the contents of different kinds of databases for quick processing.This approach paves a path toward the increase of efficient image retrieval techniques and numerous research in image indexing involving large image datasets.Image retrieval usually encounters difficulties like a)merging the diverse representations of images and their Indexing,b)the low-level visual characters and semantic characters associated with an image are indirectly proportional,and c)noisy and less accurate extraction of image information(semantic and predicted attributes).This work clearly focuses and takes the base of reverse engineering and de-normalizing concept by evaluating how data can be stored effectively.Thus,retrieval becomes straightforward and rapid.This research also deals with deep root indexing with a multidimensional approach about how images can be indexed and provides improved results in terms of good performance in query processing and the reduction of maintenance and storage cost.We focus on the schema design on a non-clustered index solution,especially cover queries.This schema provides a filter predication to make an index with a particular content of rows and an index table called filtered indexing.Finally,we include non-key columns in addition to the key columns.Experiments on two image data sets‘with and without’filtered indexing show low query cost.We compare efficiency as regards accuracy in mean average precision to measure the accuracy of retrieval with the developed coherent semantic indexing.The results show that retrieval by using deep root indexing is simple and fast.

Denoising Letter Images from Scanned Invoices Using Stacked Autoencoders

Invoice document digitization is crucial for efficient management in industries.The scanned invoice image is often noisy due to various reasons.This affects the OCR(optical character recognition)detection accuracy.In this paper,letter data obtained from images of invoices are denoised using a modified autoencoder based deep learning method.A stacked denoising autoencoder(SDAE)is implemented with two hidden layers each in encoder network and decoder network.In order to capture the most salient features of training samples,a undercomplete autoencoder is designed with non-linear encoder and decoder function.This autoencoder is regularized for denoising application using a combined loss function which considers both mean square error and binary cross entropy.A dataset consisting of 59,119 letter images,which contains both English alphabets(upper and lower case)and numbers(0 to 9)is prepared from many scanned invoices images and windows true type(.ttf)files,are used for training the neural network.Performance is analyzed in terms of Signal to Noise Ratio(SNR),Peak Signal to Noise Ratio(PSNR),Structural Similarity Index(SSIM)and Universal Image Quality Index(UQI)and compared with other filtering techniques like Nonlocal Means filter,Anisotropic diffusion filter,Gaussian filters and Mean filters.Denoising performance of proposed SDAE is compared with existing SDAE with single loss function in terms of SNR and PSNR values.Results show the superior performance of proposed SDAE method.

A Novel Adaptive Approach to Process Binary Fingerprint Images Using Directional Morphological Operations

An adaptive algorithm for removing false ridges, bridges and filling gaps in binary fingerprint images based on morphological operations is presented. A novel procedure for structuring elements design based on the specific fingerprint characteristic is described. Using the images from FVC2000 database, we have compared our method proposed here with the approach proposed by other ones. The Experimental results have demonstrated the efficiency of our method.

Research on single image super-resolution based on very deep super-resolution convolutional neural network

Single image super-resolution(SISR)is a fundamentally challenging problem because a low-resolution(LR)image can correspond to a set of high-resolution(HR)images,while most are not expected.Recently,SISR can be achieved by a deep learning-based method.By constructing a very deep super-resolution convolutional neural network(VDSRCNN),the LR images can be improved to HR images.This study mainly achieves two objectives:image super-resolution(ISR)and deblurring the image from VDSRCNN.Firstly,by analyzing ISR,we modify different training parameters to test the performance of VDSRCNN.Secondly,we add the motion blurred images to the training set to optimize the performance of VDSRCNN.Finally,we use image quality indexes to evaluate the difference between the images from classical methods and VDSRCNN.The results indicate that the VDSRCNN performs better in generating HR images from LR images using the optimized VDSRCNN in a proper method.

FAST TEXT LOCATION BASED ON DISCRETE WAVELET TRANSFORM

The paper describes a texture-based fast text location scheme which operates directly in the Discrete Wavelet Transform (DWT) domain. By the distinguishing texture characteristics encoded in wavelet transform domain, the text is fast detected from complex background images stored in the compressed format such as JPEG2000 without full decompress. Compared with some traditional character location methods, the proposed scheme has the advantages of low computational cost, robust to size and font of characters and high accuracy. Preliminary experimental results show that the proposed scheme is efficient and effective.

Simulation and Experimental Method for Microwave Oven

The simulation software, HFSS （high frequency structure simulator）, is introduced in microwave oven design. In the cold test, a network analyzer is used to measure the reflection coefficient （S11） of the cavity under empty and loaded states over the frequency range from 2.448 GHz to 2.468 GHz. In the hot test, a piece of wet thermal paper and an infrared thermal imaging camera are used to measure the electric field distributions on the mica and turntable. In the cold test, the simulation agrees well with the experiment no matter in empty state or loaded state. In the hot test, the simulation agrees well with the experiment in general in empty state and approximately in loaded state. The little difference in both cold and hot test may be due to that the model in simulation is not absolutely identical with that in experiment or the inadequate precision of infrared thermal imaging camera.

Effects of mesh style and grid convergence on numerical simulation accuracy of centrifugal pump

In order to evaluate the effects of mesh generation techniques and grid convergence on pump performance in centrifugal pump model, three widely used mesh styles including structured hexahedral, unstructured tetrahedral and hybrid prismatic/tetrahedral meshes were generated for a centrifugal pump model. And quantitative grid convergence was assessed based on a grid convergence index(GCI), which accounts for the degree of grid refinement. The structured, unstructured or hybrid meshes are found to have certain difference for velocity distributions in impeller with the change of grid cell number. And the simulation results have errors to different degrees compared with experimental data. The GCI-value for structured meshes calculated is lower than that for the unstructured and hybrid meshes. Meanwhile, the structured meshes are observed to get more vortexes in impeller passage.Nevertheless, the hybrid meshes are found to have larger low-velocity area at outlet and more secondary vortexes at a specified location than structured meshes and unstructured meshes.

Monitoring Soil Salt Content Using HJ-1A Hyperspectral Data: A Case Study of Coastal Areas in Rudong County, Eastern China

Hyperspectral data are an important source for monitoring soil salt content on a large scale. However, in previous studies, barriers such as interference due to the presence of vegetation restricted the precision of mapping soil salt content. This study tested a new method for predicting soil salt content with improved precision by using Chinese hyperspectral data, Huan Jing-Hyper Spectral Imager(HJ-HSI), in the coastal area of Rudong County, Eastern China. The vegetation-covered area and coastal bare flat area were distinguished by using the normalized differential vegetation index at the band length of 705 nm(NDVI705). The soil salt content of each area was predicted by various algorithms. A Normal Soil Salt Content Response Index(NSSRI) was constructed from continuum-removed reflectance(CR-reflectance) at wavelengths of 908.95 nm and 687.41 nm to predict the soil salt content in the coastal bare flat area(NDVI705 < 0.2). The soil adjusted salinity index(SAVI) was applied to predict the soil salt content in the vegetation-covered area(NDVI705 ≥ 0.2). The results demonstrate that 1) the new method significantly improves the accuracy of soil salt content mapping(R2 = 0.6396, RMSE = 0.3591), and 2) HJ-HSI data can be used to map soil salt content precisely and are suitable for monitoring soil salt content on a large scale.

Hiding Information into Palette-Based Image

After pointing out the weakness of the known palette-based image information hiding by palette matrix, a new spacial effective robust information hiding algorithm is proposed, which can resist the operation of ‘select all’, ‘copy’, ‘paste’ from cover to original, and can resist gently modification the palette matrix, and can resist the image format changed between true color image and palette-based image. The hiding capacity can reach 25% of the number of pixel index matrix. Due to the advisement of information hiding security an update algorithm is proposed at the end of the paper, with the capacity reduced and vision effect increased.

Experimental far-field imaging properties of high refractive index microsphere lens

The far-field imaging properties of a high index microsphere lens spatially separated from the object are experimentally studied. Our experimental results show that, for a Blu-ray disk whose spacing is 300 nm, the high index microsphere lens also can discern the patterns of the object sample when the distance between the lens and the object is up to 5.4 μm. When the distance is increased from 0 to 5.4 μm, for the microsphere lens with a diameter of 24 μm, the lateral magnification increases from 3.5× to 5.5×, while the field of view decreases from 5.1 to 3.0 μm. By varying the distance between the lens and the object, the optical image can be optimized. We also indicate that the far-field imaging capability of a high index microsphere lens is dependent on the electromagnetic field intensityprofile of the photonic nanojet under different positions of the microsphere lens.