Abstract: A calibration concentration selection method includes steps of: using a gas sensor array to obtain a concentration variation sequence and a response variation sequence of the gas mixture; constructing and training an AE-BP model; constructing VAE and identically distributing the response variation sequence; inputting the identically distributed response variation sequence into the trained AE-BP model to output a predicted concentration variation sequence; and then normalizing the predicted concentration variation sequence to generate a target concentration variation sequence; sorting a target concentration variation sequence and calculating a response gradient sequence; processing the response gradient sequence for obtaining a corresponding smoothed gradient sequence; if the spike is greater than a preset hyperparameter, finding a large gradient concentration interval; and selecting concentration test points by random uniform sampling according to weights; and selecting concentration test points from all othe

Abstract: Examples described herein relate to offload reliable transport management to a network interface device and store packets to be resent, based on received packet receipt acknowledgements (ACKs), into one or more kernel space queues that are also accessible in user space.

Abstract: The present disclosure provides a method for simultaneously measuring the value of forsterite and trace elements in olivine, comprising the following steps: Step S1: selecting samples, wherein the samples are olivine samples; Step S2: placing the samples in a sample chamber of LA-ICP-MS, and adjusting the position of the samples in the optical axis direction so that the laser beam is well focused; Step S3: optimizing the instrument to make the signal-to-noise ratio of 57Fe be the best; Step S4: adopting LA-ICP-MS peak hopping mode and receiving all the mass peaks of the samples by single electron multiplier (SEM). The present disclosure overcomes the disadvantages of long test cycle and high test cost in the prior art.

Abstract: The present disclosure provides a method for simultaneously measuring the value of forsterite and trace elements in olivine, comprising the following steps: Step S1: selecting samples, wherein the samples are olivine samples; Step S2: placing the samples in a sample chamber of LA-ICP-MS, and adjusting the position of the samples in the optical axis direction so that the laser beam is well focused; Step S3: optimizing the instrument to make the signal-to-noise ratio of 57Fe be the best; Step S4: adopting LA-ICP-MS peak hopping mode and receiving all the mass peaks of the samples by single electron multiplier (SEM). The present disclosure overcomes the disadvantages of long test cycle and high test cost in the prior art.

Abstract: Embodiments of the present disclosure provide a storage system, including a distribution apparatus and a storage device; the storage device includes at least two storage controllers and multiple storage units, where each storage unit is configured with any one of the at least two storage controllers as a home storage controller; and the distribution apparatus includes a front-end interface and at least two back-end interfaces, where the front-end interface is configured to connect to a host device, and each back-end interface is connected to each storage controller in a one-to-one correspondence manner.

Abstract: Embodiments of the present disclosure provide a storage system, including a distribution apparatus and a storage device; the storage device includes at least two storage controllers and multiple storage units, where each storage unit is configured with any one of the at least two storage controllers as a home storage controller; and the distribution apparatus includes a front-end interface and at least two back-end interfaces, where the front-end interface is configured to connect to a host device, and each back-end interface is connected to each storage controller in a one-to-one correspondence manner. According to the technical solutions provided in the present disclosure, the distribution apparatus parses an IO read/write instruction, so that the IO read/write instruction can be accurately sent to the home storage controller, which avoids forwarding the IO read/write instruction between the storage controllers, thereby improving IO processing efficiency of the storage system.

Abstract: Embodiments of the present disclosure provide a storage system, including a distribution apparatus and a storage device; the storage device includes at least two storage controllers and multiple storage units, where each storage unit is configured with any one of the at least two storage controllers as a home storage controller; and the distribution apparatus includes a front-end interface and at least two back-end interfaces, where the front-end interface is configured to connect to a host device, and each back-end interface is connected to each storage controller in a one-to-one correspondence manner. According to the technical solutions provided in the present disclosure, the distribution apparatus parses an IO read/write instruction, so that the IO read/write instruction can be accurately sent to the home storage controller, which avoids forwarding the IO read/write instruction between the storage controllers, thereby improving IO processing efficiency of the storage system.

Abstract: In an extracting step, the extracting portion obtains a linked component composed of a plurality of mutually linking pixels from a character string region composed of a plurality of characters, and extracts section elements from the character string region, the section elements each being surrounded by a circumscribing figure circumscribing to the linked component. In the first altering step, the first altering portion combines section elements at least having a mutually overlapping part among the extracted section elements so as to prepare a new section element. In the first selecting step, the first selecting portion determines a reference size in advance and selects section elements having a size greater than the reference size, from among the section elements altered in the first altering step.

Abstract: A preprocessing section binarizes input image data and calculates a total black pixel ratio. A feature extracting section detects connected components included in the binary image data and detects circumscribing bounding boxes of the connected components. Predetermined connected components are removed from all of the connected components based on the sizes of the detected circumscribing bounding boxes and bounding box black pixel ratios. By using the connected components that remain after removing the unnecessary connected components, a histogram is generated by specifying the sizes of the circumscribing bounding boxes as classes and numbers of the connected components as the frequencies of occurrence. A determining section determines whether the input image data is document image data or non-document image data based on information related to the generated histogram and the total black pixel ratio.

Abstract: An image document processing device extracts a character sequence image having M number of characters in an image document, divides the image into individual character images, extracts features of the individual character images, and based on the features, selects N (N is an integer more than 1) character images in the order of degree of matching from a font-feature dictionary for storing features of all character images according to fonts, and generates an M×N index matrix for the extracted character sequence. In searching, the device searches an index-information storage section with respect to each search character included in a search keyword in an input search expression, and extracts an image document including an index matrix including the search keyword. This provides an image document processing device and an image document processing method each allowing indexing not requiring user's operation and each allowing highly precise searching without OCR recognition.

Abstract: A headline-region initial processing section clips a headline-region image in an image document, divides the image into individual character images, and extracts features of the individual character images. Based on the features, a candidate-character-sequence generating section selects N (N is an integer more than 1) character images as candidate characters in the order of degree of matching from a font-feature dictionary for storing features of individual character images, and generates M×N index matrix where M is the number of characters in an extracted character sequence. Based on the index matrix, a document-name generating section generates a meaningful document name according to the image document. An image-document-DB management section manages accumulated image documents using the document name.

Abstract: It is an object of the present invention to provide a method and apparatus for locating a keyword of a speech and a speech recognition system. The method includes the steps of: by extracting feature parameters from frames constituting the recognition target speech, forming a feature parameter vector sequence that represents the recognition target speech; by normalizing of the feature parameter vector sequence with use of a codebook containing a plurality of codebook vectors, obtaining a feature trace of the recognition target speech in a vector space; and specifying the position of a keyword by matching prestored keyword template traces with the feature trace. According to the present invention, a keyword template trace and a feature space trace of a recognition target speech are drawn in accordance with an identical codebook. This causes resampling to be unnecessary in performing linear movement matching of speech wave frames having similar phonological feature structures.

Abstract: An image of a character string composed of M pieces of characters is clipped from a document image, and the image is divided into separate characters. Image features of each character image are extracted. Based on the image features, N (N>1, integer) pieces of character images in descending order of degree of similarity are selected as candidate characters, from a character image feature dictionary which stores the image features of character image in units of character, and a first index matrix of M×N cells is prepared. A candidate character string composed of a plurality of candidate characters constituting a first column of the first index matrix, is subjected to a lexical analysis according to a language model, and whereby a second index matrix having a character string which makes sense is prepared. In the language model, statistics are taken and then, the lexical analysis is performed.

Abstract: A preprocessing section binarizes input image data and calculates a total black pixel ratio. A feature extracting section detects connected components contained in the binarized image data and detects circumscribing bounding boxes that circumscribe these connected components, respectively. Based on sizes of the circumscribing bounding boxes detected and numbers of black pixels contained therein, predetermined connected components are removed. A determining section generates an edge map by using the residual connected components, and performs two-dimensional fast Fourier transform thereon to generate spectral data. The determining section performs two-dimensional fast Fourier transform on template images to generate spectral data. The determining section determines, based on these pieces of spectral data, whether or not a circular shape is contained in the input image data.

Abstract: An image of a character string composed of M pieces of characters is clipped from a document image, and the image is divided character by character, and image features of each character image are extracted. On the basis of the image features, N (N>1, integer) pieces of character images in descending order of degree of similarity are selected as candidate characters from a character image feature dictionary which stores the image features of character image in units of character, and the first index matrix of M×N cells is prepared. A candidate character string composed of a plurality of candidate characters constituting the first column of the first index matrix, is subjected to a lexical analysis according to a predetermined language model, whereby a second index matrix adjusted into a character string which makes sense is prepared to be utilized for searching.

Abstract: A handwriting recognition method and a handwriting recognition device are provided to recognize a character sequence continuously inputted by a user for convenience. The present method comprises steps of calculating various features of the inputted character sequence which include single character recognition accuracy features and space geometry features of different stroke combinations in the inputted character sequence, calculating segmentation reliabilities of respective stroke combinations in different segmented patterns by using a probabilistic model in which coefficients of the probabilistic model are estimated by a parameter estimation method through sample trainings, recognizing characters in different writing patterns by using a multiple-template matching method when performing single character recognition of the stroke combinations, searching for the best segmentation path and conducting post-processing to optimize the recognition results.

Abstract: Feature vectors used in discrimination of images include information on feature blocks of images in an image-document retrieving apparatus of the present invention. Text areas of a page image document are combined to form rectangular images. On the basis of information on the rectangular images that are extracted, a geometric structure of the page is analyzed, the page image document is divided into plural blocks, and then a plurality of feature blocks describing features of the page document image are selected from the plural blocks. The feature vectors are constituted of information on the feature blocks thus selected. This makes it possible to provide an image-document retrieving apparatus and a method of retrieving image documents, by which retrieval of image documents containing mainly text and a graphic is improvable in accuracy.

Abstract: It is an object of the present invention to provide a method and apparatus for locating a keyword of a speech and a speech recognition system. The method includes the steps of: by extracting feature parameters from frames constituting the recognition target speech, forming a feature parameter vector sequence that represents the recognition target speech; by normalizing of the feature parameter vector sequence with use of a codebook containing a plurality of codebook vectors, obtaining a feature trace of the recognition target speech in a vector space; and specifying the position of a keyword by matching prestored keyword template traces with the feature trace. According to the present invention, a keyword template trace and a feature space trace of a target speech are drawn in accordance with an identical codebook. This causes resampling to be unnecessary in performing linear movement matching of speech wave frames having similar phonological feature structures.

Abstract: A preprocessing section binarizes input image data and calculates a total black pixel ratio. A feature extracting section detects connected components contained in the binarized image data and detects circumscribing bounding boxes that circumscribe these connected components, respectively. Based on sizes of the circumscribing bounding boxes detected and numbers of black pixels contained therein, predetermined connected components are removed. A determining section generates an edge map by using the residual connected components, and performs two-dimensional fast Fourier transform thereon to generate spectral data. The determining section performs two-dimensional fast Fourier transform on template images to generate spectral data. The determining section determines, based on these pieces of spectral data, whether or not a circular shape is contained in the input image data.

Abstract: A preprocessing section binarizes input image data and calculates a total black pixel ratio. A feature extracting section detects connected components included in the binary image data and detects circumscribing bounding boxes of the connected components. Predetermined connected components are removed from all of the connected components based on the sizes of the detected circumscribing bounding boxes and bounding box black pixel ratios. By using the connected components that remain after removing the unnecessary connected components, a histogram is generated by specifying the sizes of the circumscribing bounding boxes as classes and numbers of the connected components as the frequencies of occurrence. A determining section determines whether the input image data is document image data or non-document image data based on information related to the generated histogram and the total black pixel ratio.