34results about How to "Eliminate distracting information" patented technology

The invention relates to the technical field of recognizing lane lines, particularly to a method for effectively judging and recognizing vehicle travelling conditions and a device thereof. The method comprises the following steps: obtaining external parameters through camera calibration; obtaining lane lines of an image frame; calculating entrance angles between a vehicle and the lane lines in real time, using lateral distances between wheels and the left and right lane lines, the entrance angles and self vehicle speed information to judge the vehicle travelling condition, and giving a danger early warning. The device comprises a judging and recognizing parameter obtaining unit, a judging and recognizing parameter calculating unit and a vehicle travelling condition judging and recognizing unit, which are successively connected. The method and the device have the beneficial effects that: 1. the lane lines can be accurately recognized, and detection rates of the lane lines in severe environments are high; 2. the calculating complexity is low, and the real-time performance is high; 3. the entrance angles, the lateral distances and lateral speeds are used to provide a self-adapting early warning of lane deviation, and the interference from error signals of early warnings or delay warnings can be avoided.

The invention discloses a method for sub-surface retrieval of MOOC courses based on ElasticSearch. The technical scheme has the following steps of firstly acquiring the MOOC courses metadata, parsing and then storing; creating an index of the MOOC courses metadata in a ElasticSearch cluster; adding data of a plurality of fields to the fields to be retrieved, achieving the multi-field search, and setting different weights for the different fields; re-setting a plurality of sub-surfaces, screening and filtering the MOOC courses from the different sub-surfaces, and allowing users to create their own search path to achieve the sub-surface search, and setting a sort. The sorting according to the specific sub-face based on the MOOC courses search results makes the MOOC courses with the expected attributes as far as possible near the front, and the search results sort is completed; finally, a search service interface is set up, the MOOC courses retrieval services are provided. After users provide a retrieval keyword and the retrieval sub-surface information, the retrieval service interface returns the MOOC courses information lists and the courses sub-face aggregation results to the users.

The embodiment of the invention discloses a tissue component non-invasive detection device and system, and wearable equipment. The device comprises an incidence optical fiber, a detection module and aprocessing module, wherein the detection module is in communication connection with the processing module. The incidence optical fiber is used for emitting incident lights with various preset wavelengths to a detected part; the detection module is used for obtaining intensity values of lights emitted from the surface of the detected part at preset positions according to each preset wavelength andsending each light intensity value to the processing module; and the processing module is used for determining the concentration of tissue components to be detected according to the light intensity value under each preset wavelength. According to the embodiment of the invention, the intensity values of the lights emitted from the detected part can be directly processed, so that light loss is reduced, and detection efficiency is further improved. As the volume of the detection device is greatly reduced, the device is easy to wear and fix on a part to be detected, and stability of detection conditions can be ensured.

The invention discloses a lane line detection method. The method comprises the following steps: preprocessing an image collected in real time; eliminating interference information on the lane line byadopting a morphological top hat operation for the preprocessed image; performing binarization processing on the image after eliminating the interference information on the lane line to obtain an initial lane line communication region; dividing the initial lane line communication region into an initial left lane line communication region and an initial right line communication region according tothe direction of the initial lane line communication region, and performing linear fitting on the initial left lane line communication region and the initial right line communication region by adopting RANSAC; and respectively performing boundary extraction on the optimal initial left lane line communication region and / or the optimal initial right line communication region after the RANSAC classification, performing B-spline curve interpolation fitting lane line processing on the extracted boundaries to obtain the final left lane line and / or the final right lane line.

The invention discloses a video caption positioning method, electronic equipment and a computer storage medium. The video caption positioning method comprises the following steps: acquiring all imageframes of a video, and performing text detection on all the image frames to obtain a first textbox set of all the image frames, the first textbox set comprising a textbox of each image frame; traversing the first textbox set, and obtaining first similarity of textboxes in every two image frames in a first direction; constructing a first graph network about the first textbox set based on the plurality of first similarities; and clustering the first graph network to obtain a plurality of first sub-networks, and extracting a textbox of the video caption from the first sub-networks of which the node number meets a first preset condition. Through the video caption positioning method, the video caption positioning accuracy can be improved, and other interference information in the video is effectively eliminated.

The invention relates to a two-dimensional code decoding method and a terminal, and belongs to the field of data processing. The method comprises the steps of by obtaining a grayscale image of a two-dimensional code, obtaining a gray histogram corresponding to the gray image; Counting a half of the total number of pixels of the grayscale image to obtain a target number; Obtaining a gray thresholdvalue according to a preset formula; Wherein the preset formula is specifically shown in the description; Wherein histGram [i] represents the number of pixels with the gray value of i in the gray histogram, S represents the target number, dif (t) represents the difference between the number of pixels with the gray value of not more than t and the target number, and T represents the gray threshold;Binarizing the gray level image according to the gray level threshold value to obtain a binarized image; And calling a bar code decoding algorithm, and trying to decode the binary image. The two-dimensional code decoding success rate of non-uniform ink jet is improved.

The embodiment of the invention discloses a tissue component non-invasive detection device, a tissue component non-invasive detection system and wearable equipment. The device comprises a light sourcemodule, a detection module and a processing module, wherein the detection module is in communication connection with the processing module; the light source module is used for respectively emitting incident light under each preset wavelength to a measured part; the detection module is used for obtaining a light intensity value emitted from the surface of a detected part at each preset position according to each preset wavelength and sending each light intensity value to the processing module, and each preset position is away from the center of the incident light at a corresponding preset distance; and the processing module is used for determining the concentration of a tissue component to be detected according to each light intensity value under each preset wavelength. According to the embodiment of the invention, the light intensity value emitted from the detected part can be directly processed, so that the light loss is reduced, and the detection efficiency is improved. Due to the fact that the size of the detection device is greatly reduced, the device is easy to wear and fix on the detected part, and the stability of detection conditions can be guaranteed.

The invention discloses a deep learning training recognition method, and the method comprises the following steps: carrying out the first target positioning training of a sample image through employing a deep learning model, extracting a target from the sample image, calibrating the target, and generating a target positioning model through training; classifying the samples to generate a classifiedsample set; identifying targets in the classification sample set by using the target positioning model trained for the first time, taking out an identified target area image, and storing the identified target area image as a new classification set sample; performing second training on the new classification set samples and the classification sample set by adopting a network model for second training to generate a classification recognition model. In the recognition process, a target is recognized by the target positioning model, and then classification recognition is performed through the classification recognition model. According to the training recognition method for deep learning, the recognition method of first positioning and then classification is adopted, and the recognition rateof deep learning can be effectively improved.

The invention discloses a pseudosciaena crocea storage time detection method. The method includes: making the laser emitted by a laser array irradiate a sample according to different angles Aj in order, collecting the light reflected by the sample by an optical fiber probe and inputting the collected light into a spectrometer so as to obtain n detection signals corresponding to the wavelengths of n lasers; inputting x(t) into a stochastic resonance system model by a computer to make the stochastic resonance system model undergo stochastic resonance; and drawing an output signal noise ratio curve of the stochastic resonance system model by the computer, and conducting calculation by the computer to obtain a predicted pseudosciaena crocea storage time. With the characteristics of fast detection speed, high accuracy and good economical efficiency, the method provided by the invention eliminates the interference information of useless band stray light, and the sample storage time is better differentiated.

The invention provides a picture optimization method. The picture optimization method comprises the steps of obtaining a target picture and a plurality of corresponding reference pictures; Dividing the target picture into a plurality of target picture alignment regions according to a set region size, the adjacent target picture alignment regions having an overlapping region; On the basis of the pixel gray scales of the target picture and the reference picture, obtaining each target picture alignment region in the target picture, a corresponding reference picture alignment region in each reference picture and the similarity of the corresponding reference picture alignment region; And based on the similarity, superposing and fusing the target picture alignment areas corresponding to the target picture by using the reference picture alignment areas of the plurality of reference pictures so as to perform noise reduction operation on the target picture.

The invention discloses a picture display method and a picture display device and belongs to the field of internet. The picture display method includes acquiring page data of a web page; stopping rendering and stopping displaying other page content except for pictures in the web page according to the page data of the web page; rendering and displaying the pictures in the web page. The picture display device comprises a first acquiring module, a first stopping module and a first display module. The picture display method and the picture display device have the advantages that other interference information except for the pictures is eliminated, and user's picture viewing efficiency is improved.

The invention discloses a detection device based on a laser array and a large yellow croaker storage time detection method. Laser emitted by the laser array first irradiates a sample, light reflected by the sample is collected by an optical-fiber probe and then enters a spectrograph, the spectrograph obtains n detection signals corresponding to wavelengths of n lasers, a detection signal x(t) conforming to the requirement is obtained by adjusting the weight of each detection signal, the x(t) is inputted to a random resonance system model which is prestored in a computer, so that the random resonance system model has random resonance, and the predicted large yellow croaker storage time can be calculated through the computer. The detection device and the large yellow croaker storage time detection method have the characteristics that the detection speed is high, the accuracy is high, and the economic performance is good; interference information of useless waveband parasitic light is eliminated, and thus the storage time of the sample can be effectively distinguished.

The invention discloses a laser array based detection device and a pseudosciaena polyactis storage time detection method. The method includes: firstly making the laser emitted by a laser array irradiate a sample, collecting the light reflected by the sample by an optical fiber probe and inputting the collected light into a spectrometer so as to obtain n detection signals corresponding to the wavelengths of n lasers; then adjusting the weight of each detection signal to obtain a satisfactory detection signal x(t), inputting the x(t) into a stochastic resonance system model prestored in a computer to make the stochastic resonance system model undergo stochastic resonance, then drawing an output signal noise ratio curve of the stochastic resonance system model by the computer, and conducting calculation by the computer to obtain a predicted pseudosciaena polyactis storage Time. With the characteristics of fast detection speed, high accuracy and good economical efficiency, the device and the method provided by the invention eliminates the interference information of useless band stray light, and the sample storage time is better differentiated.

The invention discloses an image information extraction method and device, and relates to the technical field of computers. A specific embodiment of the method comprises the steps of performing text detection and recognition on a target image to obtain a text line of the target image and character information of the text line; matching the character information with a set pre-printed character template to determine the position of the pre-printed character information in the text line; performing image segmentation on the text line image corresponding to the position of the pre-printed character information to locate the pixel position of the pre-printed character information, and extracting pre-printed character color information at the pixel position; and using the background color information of the target image to replace the pre-printed character color information, and performing character detection and recognition on the target image after color replacement to obtain printed character information. According to the method, the pre-printed character color information is extracted, after the background color is used for replacing the pre-printed character color information, character detection and recognition are carried out, the printed character information is obtained, and the integrity and accuracy of gland area information recognition are improved.

The invention discloses a pseudosciaena polyactis storage time detection method. The method includes: firstly making the laser emitted by a laser array irradiate p samples in order, collecting the light reflected by the samples by an optical fiber probe and inputting the collected light into a spectrometer so as to obtain n detection signals corresponding to the wavelengths of n lasers; then adjusting the weight of each detection signal to obtain the satisfactory detection signal x(t), inputting the x(t) into a stochastic resonance system model to make the stochastic resonance system model undergo stochastic resonance, drawing an output signal noise ratio curve of the stochastic resonance system model by the computer, and conducting calculation by the computer to obtain K, introducing K into a storage time prediction model stored in the computer, and conducting calculation by the computer to obtain a predicted pseudosciaena polyactis storage Time. With the characteristics of fast detection speed, high accuracy and good economical efficiency, the method provided by the invention eliminates the interference information of useless band stray light, and the sample storage time is better differentiated.

The invention relates to an automatic identification method for pipeline fracture. The automatic identification method comprises the steps of collecting a pipeline image, converting the pipeline imageinto a grayscale image, adopting a central block image method, determining the center of a positioning pipe diameter, removing a water flow area, extracting a binary image, marking a fracture position and the like. The method can effectively eliminate interference information of the pipe diameter center, automatically screen out images with pipeline cracks, mark the positions where the cracks exist, and achieve efficient identification of pipeline crack diseases. Compared with the traditional rupture disease manual drawing method, the method has the advantages of objectivity, repeatability, consistency and the like; compared with a direct gray threshold method, the method has the advantage of low false identification rate.

The invention discloses an affine projection decoding method of layered algebra space-time codes in a downlink multi-user multiple-input-multiple-output system. The method includes the following steps: step 1. The received signals Y, channel matrixes H and noise matrixes W are respectively written in the column vector form; step 2. the channel matrixes are decomposed in different forms; step 3. polar decomposition is carried out on H11 and H12; step 4. Affine projection is carried out, namely, y1 is projected to U1 along V1; step 5. P1H is multiplied by R (y1), and repeat step 3, step 4 and step 5; step 6. the signal information received by the first user is decoded through maximum likelihood decoding.

The invention relates to a sand reservoir prediction method and device based on seismic attributes, and belongs to the technical field of geophysical exploration. The method comprises extracting the seismic attributes of respective wavelet time window sections of the top and the bottom of a target sand body through a three-dimensional seismic top and bottom time window dividing method; selecting the seismic attributes with the highest degree of conformity with a drilled sand body based on the top and bottom; separately drawing the sedimentary facies corresponding to the top and bottom seismicattribute maps; comparing the changes of the top and bottom sedimentary facies to achieve the prediction of the sand reservoir. The time window division method of the invention can effectively removethe interference information, and can further analyze the sand change according to the seismic attributes extracted by the divided time windows, thereby improving the accuracy of identifying the sandreservoir.

The invention provides an airport remote sensing image detection method based on learning, and the method comprises the steps: carrying out the Canny edge detection of a remote sensing image, removingthe noise interference in the remote sensing image, and obtaining the specific edge information of the remote sensing image; extracting the initial coordinate value of the longest straight line segment after edge detection of the remote sensing image by using Hough transform, and calculating to obtain the length of the longest straight line; and finally, carrying out learning classification prediction by applying a support vector machine, cascading the coordinate value of the starting point of the longest straight line segment with the longest length value to obtain an enhanced characteristicquantity, and inputting the enhanced characteristic quantity into the support vector machine for full learning. Therefore, the complex remote sensing image classification problem is simplified, a large amount of interference information in airport remote sensing image detection is eliminated, the powerful binary classification capacity of the support vector machine is utilized, parameters do notneed to be modified too much, the method is simple and easy under the same condition, and the 96.5% detection accuracy is achieved.

The invention relates to the technical field of video processing, and discloses a method for video subtitle recognition, which comprises the following steps of: performing character recognition on a plurality of video frames to obtain all textboxes in each video frame; determining a plurality of textbox sets according to the occurrence frequency of the textbox height; calculating a textbox width mean square error of each textbox set; and determining the subtitles according to the width mean square error of each textbox. Since the heights of the textboxes in the video are the same, but the widths of the textboxes in different video frames are different, the textboxes of the video subtitles can be accurately identified through the heights and the widths of the textboxes. When different types of video subtitles are identified, information such as video sizes and subtitle positions does not need to be considered, and subtitle classification does not need to be carried out through manual annotation. The invention further discloses a video subtitle recognition device, electronic equipment and a storage medium.

The invention discloses a decoding method of a cyclic space-time code based on affine projection in a downlink multi-user multi-input multi-output system which comprises a first step of decomposing a channel matrix into a formula described in the specification, performing polar decomposition on two symbols described in the specification and obtaining a formula 2described in the specification, wherein P (Q) is unitary matrix of the NM*1 (NM*(M-1)), a C1 (C2) is a positive definite matrix of 1*1 ((M-1)* (M-1)), defining a U to be a subspace formed by column vectors of the P, wherein the V is a subspace formed by column vectors of the Q, a second step of projecting a matrix described in the specification on the U along the V and obtaining an affine projection of the matrix, wherein the affine projection is a formula 3 described in the specification, substituting the matrix to the formula 3, carrying out simplification and obtaining formula 4 described in the specification, wherein another symbol is equal to a formula 6, a third step of using a P<H> for multiplying the formula 4 to obtain a formula 5 described in the specification, a fourth step of obtaining information of a first user through maximum likelihood decoding. The decoding method of the cyclic space-time code based on the affine projection has the advantages that under the condition that a sending end doesn't know channel status information, an affine projection method is used and interference information among multiple users in a downlink multi-user multiple input multiple output system can be removed, through the utilization of the cyclic time-space code, information of each user is enabled to pass through all information channels, and therefore spatial diversity is ensured, and information of sending signals of a sending end is recovered accurately.

The invention relates to a two-dimensional code decoding method and a terminal, belonging to the field of data processing. The present invention obtains the grayscale image of the two-dimensional code; obtains the grayscale histogram corresponding to the grayscale image; counts half of the total number of pixels in the grayscale image to obtain the target number; obtains the grayscale according to the preset formula degree threshold; the preset formula is specifically: T={t|min(dif(t))}; wherein, histGram[i] represents the number of pixels whose gray value is i in the gray histogram, S represents the number of targets, dif (t) represents the difference between the number of pixels whose gray value is not greater than t and the number of targets, and T represents the gray threshold; The above grayscale image is obtained to obtain a binary image; the barcode decoding algorithm is called to try to decode the binary image. Realize the improvement of the success rate of two-dimensional code decoding with inkjet unevenness.

The internal error calibration compensation system and calibration compensation method of the photon interference imaging system relate to the field of optical interference imaging technology, and solve the problems that various errors exist in the devices of the existing interference imaging system, which in turn affect the performance of the imaging system, including calibration of the light source and phase adjustment. device, a PIC chip to be tested, a light intensity detection device, and a fiber coupler; one end of the fiber coupler is connected to a calibration light source, and the calibration light source is divided into two outputs, and each of the two output ports at the other end of the fiber coupler is connected to a phase The input port of the adjustment device is connected, and the output ports of the two phase adjustment devices are respectively connected to the two input ports of the PIC chip to be tested; the light intensity detection device is placed at the back end of the PIC chip to be tested, and is used to detect the four output ports of the PIC chip output light intensity. For the internal error of the PIC integrated chip, the invention effectively corrects the internal error by establishing an error calibration compensation model, so that the system can achieve better imaging effect. Simplify system operation.

The invention discloses an affine projection decoding method used in a downlink multiuser multi-input multi-output (MIMO) system. The method includes the steps of: S1, processing H1 and H2 with polar decomposition and obtaining H1=PC1, H2=QC2, wherein P (Q) is unitary matrix of N*k (N*(M-k)), C1 (C2) is a positive definite matrix of k*k ((M-k)*(M-k)), letting U be a subspace composed of column vectors of P and letting V be a subspace composed of column vectors of Q, S2, simplifying a formula1 of an affine projection of Y on U along V, and obtaining a formula2, S3, multiplying PH by a formula3, S4, decoding information received by the ith user through a maximum likelihood decoding. The affine projection decoding method used in a MIMO system has the advantages that the affine projection method is capable of removing interference information among the multiple users of the downlink multiuser MIMO system when a sending terminal is unaware of channel state information, and information of a signal sent by the sending terminal.

The invention relates to a sandbody reservoir prediction method and device based on seismic attributes, belonging to the technical field of geophysical exploration. The present invention extracts the seismic attributes of the time window intervals of the top and bottom subwaves of the target sand body through the division method of the three-dimensional seismic top and bottom time windows, selects the seismic attributes with the highest degree of conformity with the drilled sand body by dividing the top and bottom, and draws the top and bottom respectively. The sedimentary facies corresponding to the bottom seismic attribute map, and the changes of the top and bottom sedimentary facies are compared to realize the prediction of sand body reservoirs. The time window division method of the present invention can effectively remove interference information, further analyze sand body changes according to the seismic attributes extracted by the divided time windows, and improve the accuracy of identifying sand body reservoirs.

The present invention provides a picture optimization method, which includes: acquiring a target picture and corresponding multiple reference pictures; dividing the target picture into multiple target picture alignment areas according to the set area size, and the adjacent target picture alignment areas have overlapping Area: Based on the pixel gray scale of the target picture and the reference picture, obtain each target picture alignment area in the target picture, the corresponding reference picture alignment area in each reference picture, and the similarity with the corresponding reference picture alignment area; based on the similarity , superimposing and fusing the corresponding target picture alignment areas of the target picture by using the reference picture alignment areas of multiple reference pictures, so as to perform a noise reduction operation on the target picture.

The invention discloses a mobile detection system and method for pavement gap detection, and the mobile detection of a pavement gap is realized by additionally installing a positioning and control module, a millimeter wave radar and a real-time processing module on a system bearing vehicle. The millimeter wave radar is used for road gap detection, compared with traditional manual detection, the labor cost is reduced, and meanwhile the detection efficiency is improved. And compared with a traditional fixed optical detection system which needs to be provided with a plurality of optical systems, the detection cost is also reduced. The millimeter wave radar is adopted for detection, stable monitoring can be carried out on tiny gaps with the width smaller than 1 cm, and the stability and comprehensiveness of detection are improved. The mobile detection system is loaded on the system bearing vehicle, all-weather uninterrupted detection can be carried out, the influence of weather conditions is avoided, and the universality and the stability of the detection system are improved.

The invention relates to the technical field of recognizing lane lines, particularly to a method for effectively judging and recognizing vehicle travelling conditions and a device thereof. The method comprises the following steps: obtaining external parameters through camera calibration; obtaining lane lines of an image frame; calculating entrance angles between a vehicle and the lane lines in real time, using lateral distances between wheels and the left and right lane lines, the entrance angles and self vehicle speed information to judge the vehicle travelling condition, and giving a danger early warning. The device comprises a judging and recognizing parameter obtaining unit, a judging and recognizing parameter calculating unit and a vehicle travelling condition judging and recognizing unit, which are successively connected. The method and the device have the beneficial effects that: 1. the lane lines can be accurately recognized, and detection rates of the lane lines in severe environments are high; 2. the calculating complexity is low, and the real-time performance is high; 3. the entrance angles, the lateral distances and lateral speeds are used to provide a self-adapting early warning of lane deviation, and the interference from error signals of early warnings or delay warnings can be avoided.