40 results about "False accept rate" patented technology

The present invention includes a method of deciding whether a data set is acceptable for making a decision. A first probability partition array and a second probability partition array may be provided. One or both of the probability partition arrays may be a Copula model. A no-match zone may be established and used to calculate a false-acceptance-rate (“FAR”) and/or a false-rejection-rate (“FRR”) for the data set. The FAR and/or the FAR may be compared to desired rates. Based on the comparison, the data set may be either accepted or rejected. The invention may also be embodied as a computer readable memory device for executing the methods.

The invention is a fingerprint discriminating method based on broken design detection which belongs to fingerprint identification field. Its character lies in: it carries on broken design detection to detail points extracted with traditional method with 12 groups of best filter, each filter has a discrete value of an angle between the broken design direction and the X axis; works out the filter value under the 12 groups of broken design directions and sets a uniform floor level, acquires 12 groups of binary image, then the stripe region is displayed with rectangular with primary cell analysis method of matrix, and the region is parameterized with length, width and direction of the rectangular; all the binary images are combined together and a real broken design image is acquired, the parameterized images are combined and a broken design model is acquired in order to eliminate the false detail points. To the old people, with the same error reception rate, the fault refusing rate can be decreased to less than 10%.

A field programmable gate array (FPGA) device for diagnosing and predicting artificial circuit faults comprises an FPGA board card, wherein the FPGA board card is connected with a computer host machine in Ethernet communication mode, and upper machine software is operated in the computer host machine. The FPGA board card is used for acquisition, cache, preprocessing and back-up of accessible node voltage information and network spectrum characteristic data of an artificial circuit to be tested and packing and transmitting an original signal and artificial circuit fault information to the upper machine software. The upper machine software is used for receiving data frames sent by the FPGA board card, provides interface software for diagnosing and predicting the artificial circuit faults, simultaneously provides a delivery interface of original data and can seamlessly deliver the data to other analysis platforms of the computer host machine to perform follow-up processing, analysis and back-up. The FPGA device for diagnosing and predicting the artificial circuit faults tests and diagnoses some large-scale naval ship artificial circuits and a radio frequency identification device (RFID) filter network, the accuracy rate of fault diagnosis is larger than 95%, and the false rejection rate and the false accept rate are smaller than 3%.

The invention discloses a fingerprint identification method and a fingerprint identification device, belonging to the field of fingerprint identification. The fingerprint identification method comprises steps: whether the number of damaged pixel units in a fingerprint identification sensor reaches a predetermined threshold is detected, wherein the damaged pixel units are physically-damaged pixel units in the fingerprint identification sensor; and if the number of the damaged pixel units reaches the predetermined threshold, matching and identification on a fingerprint image acquired by the fingerprint identification sensor are abandoned. The problem that a learning function adds pixel features of the damaged pixel units in the fingerprint identification sensor to a fingerprint template, the false accept rate of the fingerprint identification sensor is thus improved, and hidden safety dangers are caused to exist in a mobile terminal is solved. The effects that after the number of damaged pixel units reaches the predetermined threshold, matching and identification on the fingerprint image are abandoned, the false accept rate of the fingerprint identification sensor is reduced, and hidden safety dangers brought by the over high false accept rate of the mobile terminal are achieved.

The invention discloses a medical image classification method based on SVM. The method is characterized by comprising the following steps that 1) medical images are stored in an original image data base, the medical images comprise digital photos and donor information; 2) a feature recognizer is used for recognizing the medical features of the medical images in the original image data base; 3) a classifier is used for recognizing diseases according to the medical features obtained by the feature recognizer, and the medical features are stored in a classification image data base according to the types of the diseases. According to the medical image classification method based on SVM, an Adaboost cascade classifier is used for training in advance, good recognition rate and false accept rate for different samples are achieved, computing speed is quick, recognition and classification of a large number of medical images can be carried out quickly, and the method can be used for a large-scale medical image data base system.

The invention provides a method for recognizing handwritten mathematical formulas and generating MathML (mathematical makeup language) based on an Android system. The method includes collecting discrete ordinate sequences of handwritten characters on a drawing board to acquire boundary information of the characters; generating an initial image in a processor, drawing the collected discrete ordinate sequences in the initial image, and cutting the image into character images containing characters only; performing gray-scale processing and binarization processing on the character images; extracting a feature value; performing dimensionality reduction on rows and columns of two-dimensional arrays respectively; performing rough classification according to the feature value, fixing to-be-recognized characters in one class, then using a BP (back propagation) neural network to performing character matching, and acquiring optimal characters; adopting trigeminal tree structure to position spatial relationships of the mathematical formulas, using a preorder algorithm of the trigeminal tree to perform order, and acquiring the MathML; and displaying the MathML on a browser. By the aid of the method for recognizing the handwritten mathematical formulas and generating the MathML based on the Android system, false accept rate can be reduced, and system performances can be improved.

The invention provides a polymorphism reconstruction and optimization method for realizing an abnormal electrocardiogram template based on big data. The polymorphism reconstruction and optimization method is characterized by comprising the following steps that firstly, parameter template data with two-dimension characteristics are built according to decomposing parameters of standard electrocardiogram waveforms, and an electrocardiogram waveform template consisting of a plurality of fine nodes is built; then, the big data of the existing and edited dynamic electrocardiogram monitor are utilized for preprocessing before comparison, i.e., waveform segmentation, and next, the gradual waveform comparison with the electrocardiogram waveform template is carried out; next, three key quantities including FRR (fault read rate), FAR (false accept rate) and TH of the value range being 0 to 1 are used for counting the success rate finally matched with the assessment in a comparison algorithm, in addition, the polymorphism reconstruction is carried out, or the template data volume optimization is carried out again.

The invention relates to a semantic recognition-oriented paper translation method and a semantic recognition-oriented translation pen device. The semantic recognition-oriented paper translation method comprises the following steps: (1) carrying out basic encoding on English characters; establishing a character encoding library, a rule library and a font library; combining and arraying the character encoded library, the rule library and the font library to form an encoding preparation library; (2) carrying out scanning recognition on paper English to be translated by using OCR (Optical Character Recognition); (3) encoding recognized character strings by using the encoding preparation library; (4) carrying out semantic processing on the encoded character strings so as to finish encoding semantic description; (5) obtaining accurately-recognized English words by using cognitive inference of OCR for recognizing words; (6) connecting the English words which are accurately recognized by the OCR with an electronic dictionary so as to realize automatic translation. Compared with the prior art, encoding, semantic processing, inference and traditional OCR are combined so that the false accepted rate caused by traditional OCR file recognition is reduced.

This invention creates numerous enrollment reference images so that when an individual presents himself to a biometric sensor, the individual's live image is compared to numerous reference images allowing to establish a very high matching threshold score and at least one reference image to surpass the matching threshold for an individual in the file. Since the acceptance threshold is set high, the chances that a false accept are significantly reduced.

In our system, by selecting numerous different photographs of the enrolled candidate, there's a greater possibility of a very high score match to the live scan. As you increase the number of biometric photos, you increase the probability of the live scan achieving a high match against one of those photos. This could be visualized as a shotgun distribution on a graph; if the live scan is the person in a reference file, that shotgun distribution will place higher compared to someone that doesn't match the biometric. All it takes is one image to cross the threshold to achieve a match. The series of biometric images comprises what can be collectively referred to as the “shotgun template.” The probability of having a live scan match against a large number of photos is greater than against a small number of photos. Our method of enrollment consists of moving the biometric multi-directionally to create numerous varied biometric image templates.

The next part of our invention is to extract a sample of the video images since the video captures the motion of moving right-left or up-down, every third fourth or some other number of frames can bet taken as a sample, instead of searching every frame. Cameras normally capture 30-60 frames a second. A lesser number would be selected such as 1-10 frames/second, to build a signature file of that individual. The signature file is converted into templates that can be searched in the biometric search engine against the live scan. The live scan is image taken of the person's biometric that is converted to a template and added to the biometric matcher. To increase the probability of a match, a similar approach could be used on the live scan; instead of a single photo of someone approaching the biometric collector, multiple photos could be take or a video could be used to create a high number of images. The challenges of today's match engines are that due to template size, comparisons of multiple images against multiple images quickly reaches physical limitations for computer systems. Recent breakthroughs in template size will allow billions of comparisons in seconds, which was previously not possible. The recent developments in very small templates through machine learning results in numbers over a million being capable of being stored in a mobile device. This new capability of matching numerous templates with this invention's ability to create a high number of templates when someone is enrolling and when a live image is taken.

The invention discloses a vena metacarpea image identifying method, comprising: collecting hand vena metacarpea images through an image acquisition device; graying color images; determining and marking an identification area according to the size of a connected region; generating a marked area into a 2DPCA characteristic database; partitioning pixels in the characteristic database; performing gauss noise reduction processing on partitioned pixels; performing weighting treatment on the projection matrix of each sub-block after blocking partition; removing the salt and pepper noise of the obtained result through weighting treatment to obtain characteristic data; and comparing the obtained characteristic data with the sample database in a sample database to obtain an identification result. The vena metacarpea image identifying method has a false reject rate less than 0.001%, and a false accept rate less than 0.00001%, possesses the characteristics of less operational data and short time, and facilitates industrialization.

The invention discloses an intelligent multi-angle knock-on type algorithm coded lock and relates to the field of the coded locks. The intelligent multi-angle knock-on type algorithm coded lock comprises a coded lock part and an intelligent control system for opening and closing the coded lock part in a three-axis six-face knock-on mode. The intelligent control system comprises a power supply part for supplying power to the whole system, a MCU processor used as a control operation core, a data storage module, a motor driving circuit, a prompt part and a sensor module. The MCU processor adopts a MCU control chip. The output end of the motor driving circuit controls a motor for opening and closing the coded lock. The MCU processor outputs a control signal to the motor driving circuit. Opening of the coded lock is achieved in a three-axis six-face knock-on mode, no key is needed completely, the code combination number is greatly increased, the false accept rate (misoperation prevention) is low, the true rejection rate is low, the knock-on characteristics of hands can be more easily recognized through the optimization sensor algorithm, the true accept and false rejection rate ability is further improved, and easy accept and stability are achieved.

The invention discloses a minimum mean distance-based dynamic time warping method. The method comprises the following steps of: 1) creating a mean distance matrix and a path length matrix for two time sequences; and 2) solving dynamic planning of a formula (as shown in the specification): 2.1) carrying out initialization, 2.2) carrying out assignment on the first line and first column of the mean distance matrix (as shown in the specification), carrying out weighted array on the first line and neighbors, in the same line, at the former column to calculate an arithmetic mean, and carrying out weighted array on the first column and the neighbors, in the same column, of the former line to calculate an arithmetic mean, 2.3) carrying out assignment on the lines except the first line and the columns except the first column of the mean distance matrix (as shown in the specification), combining three adjacent points, adjacent to the current point, of the former line and the former column and calculating arithmetic means, selecting the minimum mean of the obtained three means to serve as a mean distance value of the current point, and adding 1 to a path length value of the current point on the basis of a length value of the selected adjacent point, and 2.4) outputting a formula (as shown in the specification) and 1 (n, m). The minimum mean distance-based dynamic time warping method provided by the invention is relatively low in false accept rate.

An image quality evaluation method for an iris recognition system, using the "image clarity" index to determine the accuracy of the focus of the collected iris image, using the entropy of the gray level co-occurrence matrix as the evaluation function of image clarity, and using the "inner and outer eccentricity" The index determines the degree of eccentricity of the inner and outer edges of the iris in the collected iris image, that is, the degree of deformation of the iris texture, and the "iris visibility" index is used to judge the amount of effective iris texture information contained in the collected iris image, that is, to the recognition system For the ability to provide effective information, use the "standard score method" to combine the above three indicators to obtain the comprehensive quality index of the iris image, and select the image with relatively better quality in the collected image sequence for identification. The invention comprehensively reflects the image quality requirements of the iris recognition system, reduces time and space complexity, and lowers the false acceptance rate and false rejection rate of the system.