147 results about "Perceptual hashing" patented technology

The invention discloses a sidewalk vehicle illegal parking detection method based on target detection and semantic segmentation, and belongs to the field of deep learning and computer vision, and themethod comprises the steps: respectively training a semantic segmentation network and a target detection network at a training stage, and obtaining a semantic segmentation model and a target detectionmodel; the method comprises the following steps of: extracting an urban road background image by using a Gaussian mixture background model; the method comprises the following steps: acquiring a semantic segmentation model, performing semantic segmentation by using a semantic segmentation algorithm and the semantic segmentation model to obtain a semantic segmentation graph, judging whether a camera rotates or not by using a perceptual hash algorithm, performing vehicle detection by using a target detection algorithm and the target detection model, marking a vehicle detection frame, and finallycomparing the vehicle detection frame with the semantic segmentation graph. The problems that an existing vehicle illegal parking detection method is high in misjudgment rate and poor in real-time performance, a sidewalk illegal parking area needs to be manually calibrated, and when a camera rotates, the illegal parking area needs to be manually calibrated again, so that the camera is not suitable for rotating are solved.

The invention discloses an image perceptual hashing method based on two-sided random projection, which mainly solves the problem of quick searching of massive image data. The method comprises the following steps of (1) preprocessing original image data; (2) obtaining a two-sided random projection matrix; (3) carrying out low-rank approximating; (4) updating a low-rank matrix; (5) judging if the number of iteration times of the updated low-rank matrix is maximum or not; (6) grouping projection vectors; (7) orthogonalizing the projection vectors; (8) obtaining hashing codes; (9) obtaining a hamming distance; and (10) outputting test results. The hashing method has the advantages that the better projection vector can be obtained, the effective hashing codes can be obtained, the memory consumption is reduced, the searching time is saved, the comprehensive performance of precision ratio and recall ratio of image searching can be improved, and the hashing method is applied to the image searching services of electronic businesses and mobile terminal equipment.

An embodiment of the invention provides a method for searching matched pictures and a method and a device for searching pictures. By setting the first proportion threshold value and the second proportion threshold value, the first portion threshold value is larger than the second threshold value, the larger matching threshold value is used for matching of local characteristics, on this basis, pictures with matching proportion smaller than the first proportion threshold value and larger than the second proportion threshold value are screened by a perceptual hashing matching manner, pictures with Hamming distance smaller than the preset first distance threshold value are contained in picture matching results, on one hand, matching accuracy of the pictures is guaranteed by the aid of the larger matching threshold value, on the other hand, for pictures with matching superposition proportion of the local characteristics between the larger matching threshold value and the smaller matching threshold value, and the pictures can be screened in the perceptual hashing matching manner. On the premise of accuracy of the screened pictures, the number of the pictures in the searching results is increased.

The invention discloses a particle filter and perceptual hash-based video target tracking method which comprises the following steps: 1) selecting a target to be tracked from an original image as a target area, extracting a perceptual hash characteristic of the target area as a template feature, and initialization positioning a weight particle set; 2) performing a state coordinate prediction for the weight particle set and labeling the prediction area; 3) extracting the perceptual hash characteristic for each predicting particle area, and calculating a similarity degree of the predicting particle area and the target area; 4) according a position contribution of a distance from the target and a similarity degree refreshing particle weight of the perceptual hash characteristic, adopting a minimum variance to estimate and obtain a new target area; and and 5) according the size of the weight and a resample estimation value, obtaining the weight particle set of a next frame, and repeating from the step 2 to the step 4 till an end of all video sequences. The particle filter and perceptual hash-based video target tracking method disclosed by the invention is high in efficiency, and can realize fast and effective tracking on the premise of great illumination variance and change of the target scale.

The invention relates to a fast image recognition system and method based on perceptual hashing, and belongs to the technical field of image recognition. The fast image recognition system comprises aninput module, a size processing module, a color adjustment module, a DCT module, an equalization module, a reduction module, an acceleration processing module, a perceptual hashing calculation module, a comparison module and an output module. The advantages and the significance of the invention are that after the initial image inputted, the KMEANS algorithm is applied to perform acceleration calculation of the Hamming moment, then processing is performed by using the perceptual hashing algorithm and then comparison with all the image databases is performed so that the problems of time consumption and labor consumption and low efficiency of calculation of the single perceptual hashing algorithm can be solved and a new and more efficient algorithm for similarity search between images can beprovided.

The invention discloses a perceptual hashing-based practical and safe image forensic system and a perceptual hashing-based practical and safe image forensic method. The system comprises a certificate authority (CA), forensics computing platform (FCP) and a forensic center (FC). The method comprises the following steps that: the CA provides a certificate for a user, the FCP and the FC and distributes and manages a control key of a hash function for the FCP and the FC; the FCP calculates a characteristic value and generates a hash value of an image at a registration or forensics stage; the FC manages an image hash value from the FCP and provides retrieval and forensics services; the user submits the image to the FCP, extracts an image retrieval characteristic by using the FCP, calculates the hash value of an image segment by using a key distributed by the CA and submits the hash value to an FC database; and when the user needs an authentication image, the image is also submitted to the FCP, the retrieval characteristic is extracted to retrieve the most coincident original image from the FC, the image is segmented and the hash value is calculated by using the same key and the difference between the image and the hash value is compared so as to determine a position on which content tampering occurs. The system achieves high balanced performance in the aspects of safety, robustness, sensibility, rapid retrieval and the like.

The invention discloses an image privacy decision recommendation system and method under a social network on the basis of a privacy rule and perceptual Hash. The invention is characterized in that the system comprises a privacy object registration module, a privacy rule definition module and a privacy decision recommendation module, wherein the privacy object registration module provides a registration function of an object to be protected for a user; the privacy rule definition module defines a privacy security level for a to-be-protected objet of the user; and the privacy decision recommendation module identifies a to-be-uploaded object of the user, compares the to-be-uploaded object of the user with a protected object in a Hash fingerprint database and a privacy rule in a privacy rule base, and finally submits the privacy security level of the to-be-uploaded object to the privacy decision recommendation module of the user. The quick and convenient automatic calculation of the privacy level and decision recommendation can be realized, and the leakage risks of individuals and the families of the individual can be effectively lowered so as to greatly improve the protection safety of user individual image privacy.

The present invention discloses a fabric defect detection method based on a distance matching function and a perceptual hash algorithm. According to the method, the minimum period of a regular fabricis calculated through using the distance matching function, and the minimum period can be applied to obtaining image blocks; a repeat unit template is constructed, and the structure features of the template image block are extracted through adopting the perceptual hash algorithm; the structural features and grayscale features of image blocks of a sample to be tested are extracted, and are comparedwith the structural features and global gray average value of the template image block, so that a minimum Hamming distance feature image and a gray average value comparison feature image are obtained; and the merging and segmentation of the feature images are performed. The results show that the method of the present invention comprehensively considers the structural features and gray average value feature of the minimum period of the regular fabric, can effectively extract the defect area of the fabric and realize the defect detection of the fabric.

Disclosed are a sensing Hash value extracting method and a sensing Hash value authenticating method for voice sensing Hash authentication. According to the characteristics of LPC (linear prediction coefficients), robustness of an LPC method is improved by optimizing the LPC method and performing three steps of optimizing and blocking of the LPC and disintegrating matrix of parameters after blocking, the optimized LPC are high in robustness and calculating efficiency, and robustness of a sensing Hash sequence formed is improved as well. The sensing Hash value calculated by the optimized LPC maintains the advantages of good instantaneity of a simple LPC method while having good robustness for attacks to the voice during transmission.

The invention relates to an intelligent-texture anti-counterfeiting method based on perceptual hashing. A first step is characterized in that image characteristic extraction is performed, which comprises that (1) a perceptual hashing algorithm is used to process an image so as to obtain one visual characteristic vector V (j) of an original texture image; (2) a user uses a mobile phone to scan the texture image to be tested and upload to a server, the perceptual hashing algorithm is used to process the image to be tested and the visual characteristic vector V ' (j) of the image to be tested is acquired. A second step is characterized in that image discrimination is performed, which comprises (3) a normalized correlation coefficient NC value between the visual characteristic vector V (j) of the original texture image and the visual characteristic vector V ' (j) of the image to be tested is acquired; (4) the obtained NC value is returned to the mobile phone of the user. An experiment proves that the method of the invention possesses a strong conventional attack resistance capability and a geometric attack resistance capability. A problem of automatically discriminating the texture image is solved. An intelligent texture anti-counterfeiting technology is realized. Discrimination accuracy is high and a speed is fast.

The invention discloses a target detection method and device, and a computer storage medium. The target detection method comprises the steps of obtaining a template image; determining a target candidate region from the current frame image to be detected; calculating a first similarity between the target candidate region and the template image based on the perceptual hash feature; calculating a second similarity between the target candidate region and the template image based on the FHOG features; and making a common decision based on the first similarity and the second similarity so as to determine a current target area corresponding to the template image from the target candidate areas of the current frame image. In this way, the current target area can be accurately determined.

The invention discloses a video tamper detection method and a video tamper detection device based on time-domain perceptual hashing. The method comprises the steps of performing the same random partitioning on each frame of image in a to-be-detected video; calculating the brightness mean value of each partition by taking each frame as a unit, and counting the brightness mean value change condition of each partition on a time-domain passage for forming a time-domain feature matrix of the to-be-detected video; quantizing the time-domain feature matrix to obtain a corresponding hashing sequence; extracting a hashing sequence corresponding to an original video with the same video length; calculating the time-domain feature distance between the to-be-detected video and the original video according to the hashing sequences of the to-be-detected video and the original video; comparing the time-domain feature distance with a set threshold value, if the time-domain feature distance is smaller than the threshold value, the to-be-detected video is determined to be not tampered, and otherwise, the to-be-detected video is determined to be tampered. The video tamper detection speed and accuracy are greatly improved.

The invention provides a method and a device for searching similar images based on perceptual hash. The device comprises an input unit, a difference unit, a searching unit, a screenshot unit, a perceptual hash algorithm unit, a comparison unit, and an output unit. The device has beneficial effects that after an input original image and a screenshot image of the original image are processed by the perceptual hash algorithm, the input original image and the screenshot image are respectively compared with images in a search library, so that under the condition that the input image is the screenshot image of a certain image, the source of the original image can be found. The device and the method improve accurate rate of the perceptual hash algorithm, and provide a new thinking for searching similar images.

The invention discloses a blocking perception Hash tracking method with shadow removing. The method comprises the steps of: determining shadow areas in an image according to the distribution characteristics of a shadow image in each channel grey-scale map of a CIELAB color space; then utilizing a color constancy theory to recover pixel points in the shadow areas to a non-shadow effect; combining blocking perception Hash coding values with color self-similarity for forming a similarity measure, and carrying out matching on tracked target sub-blocks of adjacent frames based on the similarity measure; and finally, combining the above sub-blocks to obtain the regional position of the tracked target in the current frame, and realizing the tracking of the tracked target in a video. The blocking perception Hash tracking method has the advantages that according to different moving ranges and deforming degrees of human body parts, a human body target is divided into eight sub-blocks, and on this basis, and the blocking perception Hash coding method is provided to solve the problem of an existing tracking algorithm that the tracking is unsuccessful when the human body is partially or totally shielded or partially rotated and when the illumination in the shadow areas and non-shadow areas of a natural scene changes suddenly.

The invention discloses a video-acquisition-based Visual Map database establishing method and an indoor visual positioning method using a database, relates to the field of indoor positioning and navigation, and aims to solve the problems of low indoor visual positioning accuracy, high time consumption and high labor consumption of an existing method. The indoor visual positioning method using the database is characterized in that a platform carrying a video acquisition device is used for recording videos in the constant-speed linear motion process on the basis of the quickly established video-based Visual Map database; the acquired videos are processed for recording coordinate position information and image matching information of frames of the videos; in an on-line positioning stage, a system is used for roughly matching an image uploaded by a positioned user with the video-based Visual Map database by using a hash value which is obtained by calculating by using a perceptual hash algorithm, and completing the visual indoor positioning by virtue of the roughly matched frames and the uploaded image by using an SURF algorithm and a corresponding epipolar geometry algorithm. The indoor visual positioning method is applied to indoor visual positioning places.

The invention discloses a quick massive-picture deduplication method. Each of to-be-deduplicated pictures is used to generate image fingerprint information through a perceptual hash algorithm; and multiple sets of random hash mapping are employed to construct an image hash feature dictionary, and thus duplicate pictures are removed. Compared with the prior art, the method generates image fingerprints through constructing low-dimensional image features, constructs the image hash feature dictionary, thus carries out quick deduplication of the massive images, completely removes image feature comparison consuming more time, alleviates the mapping space sparse problem caused by the low-dimensional image features and local-sensitive hashes through designing multiple reasonable hash mapping, canquickly extract the image features, and position the duplicate images, also reduces feature comparison frequency of the images to 0, and greatly improves deduplication efficiency of the massive imagesin a case of ensuring accuracy.

The invention discloses a Chinese herbal medicine plant picture capturing method based on a professional term vector of the traditional Chinese medicine and pharmacy field. The Chinese herbal medicine plant picture capturing method comprises the following steps: firstly, selecting and collecting traditional Chinese medicine and pharmacy text data, text information of medicine books and dictionary entry information relevant to traditional Chinese medicine and pharmacy, captured from an encyclopaedia website; secondly, training a Word2Vec model by using the text data so as to obtain professional term vectors of the traditional Chinese medicine and pharmacy field; then, acquiring a candidate picture set and text information thereof on source web pages through an image retrieval function based on texts provided by common search engines, such ash Google and Baidu, and calculating characteristic vectors of the source web pages by using a Doc2Vec model; finally, according to the similarity of the characteristic vector of the source web page and the corresponding Chinese herbal medicine term vector, rearranging the candidate picture set, and selecting a plurality of pictures arranged at the front, and de-duplicating the picture set by using a perceptual hash algorithm so as to obtain a final picture set.

Automatic registration of FPC images based on SURF features and hash perception algorithm. The invention discloses a FPC image registration method based on SURF feature and hash perception algorithm,including: first selecting the template image and ROI area, selecting four rectangular regions within the ROI region, matching the rectangular region and the SURF feature points of the image to be registered is realized by using force matching or FLANN, checking the similarity of the rectangular region around the matched feature points by using perceptual hashing algorithm, and finding four pairsof matched feature points to generate perspective transformation matrix, and then registering the image to be registered by perspective transformation. The invention verifies the matched SURF characteristic points by using the perceptual hash algorithm, greatly improves the matching accuracy of the SURF characteristic points, simultaneously establishes a perspective transformation matrix by usingthe matched characteristic points, and completes the image registration by using the perspective transformation and obtains a better effect.

The invention discloses a cipher text speech perception hashing and retrieving scheme based on time-frequency domain trend change. A piece of speech is divided into a time domain part and a frequency domain part for extracting perception hash, the speech is encrypted by a high-efficiency chaotic XOR encryption algorithm adapting to large-scale data, and a perception hash sequence is embedded into the least significant bit of the cipher text speech by the digital watermarking technology to generate a cipher text speech library and a system perception hash table. The cipher text speech library and the perception hash table are uploaded to the cloud. During retrieval, a perception hash sequence is extracted from an index speech provided by a user, the abstract sequence is submitted to a cloud server as an index, and matching retrieval is carried out in the system hash table of the cloud. When the perception hash sequence is matched with a perception hash value in the system hash table, a cipher text speech corresponding to the hash abstract in the hash table is returned to the user, and retrieval succeeds. Rapid and accurate retrieval of an encrypted speech in the cloud is realized. According to the method, weight distinguishing is carried out, matching is carried out successively, and therefore, the matching efficiency in large-scale application is improved.

The invention relates to an image registration method and a device based on an improved SIFT and a hash algorithm. The method comprises the following steps: at least two images to be matched are respectively constructed into a Gaussian scale space; detecting a plurality of local extreme points in the Gaussian scale space; selecting a plurality of key points from the plurality of local extreme points, and respectively describing 72-dimensional features of the plurality of key points; by narrowing the neighborhood of the selected feature points, the dimension of the traditional SIFT descriptor is reduced, the computational complexity is reduced, and the matching speed is improved. Then, by increasing the cosine distance between the feature point vectors, the similarity between the two feature vectors to be matched can be calculated better and more effectively, and the matching precision can be improved. Finally, by means of mean-perceptual hashing operation to eliminate some mismatched point pairs, the invalid data in the sample data set is reduced, and the iterative search for the optimal parameter matrix in the matching point data set is more efficient.

The invention discloses a multi-country banknote serial number character recognition method based on image fingerprints and belongs to the technical field of banknote pattern recognition. The method comprises: based on a template set and a sample set of multi-country banknote serial number character images, generating training sample data by a perceptual hash algorithm; constructing a classification model based on a SVM classifier; training the classification model by using the training sample data; and recognizing the serial number character image of a banknote to be processed by using the trained classification module. The method of the invention converts a single feature in the process of recognizing the serial number character of the banknote into a plurality of features, and uses themethod based on the SVM classifier to recognize the characters, has small sample demand, fast training speed, fast recognition speed, a good anti-interference ability and can improve the fault tolerance rate during prediction.

The invention provides a malware family identification method and device and electronic equipment, and the method comprises the steps: carrying out the perception Hash processing of a binary program to be identified, and obtaining a plurality of perceptible Hash character strings, wherein the perceptual hash processing comprises dynamic sandbox operation, gray level image conversion and discrete cosine transformation; searching and matching each malicious software family in the malicious software database by taking each perceptible Hash character string as a search word to obtain a matching degree between each malicious software family and the binary program, wherein each malicious software family comprises a plurality of perceptible hash word string sequences obtained through perceptive hash processing; and taking the name of the malware family corresponding to the maximum matching degree in the matching degrees exceeding the preset matching degree threshold value as an identificationresult corresponding to the binary program. According to the invention, when the dynamic malware detection engine detects malware, the accurate name of the malware family to which the malware belongscan be given.

The invention relates to the field of video copyright protection, and discloses a video to-be-retrieved positioning method applied to video copyright protection, which comprises the following steps: acquiring a plurality of key frame images of each video copyright data in a video copyright data set and a plurality of key frame images of to-be-retrieved video data; extracting a feature vector of the key frame image by using a deep convolutional neural network model; calculating perceptual hash values of a plurality of key frame images of the to-be-retrieved video data and perceptual hash valuesof a plurality of key frame images of each video copyright data; constructing a to-be-retrieved system added with a clustering algorithm and an inverted index; and performing quick retrieval and positioning on the to-be-retrieved video data. According to the method, the infringement fragment or the infringement picture can be accurately positioned, the identification of possible attack resistingmeans of the infringement video is enhanced, the robustness of the model is enhanced, the retrieval efficiency under mass data is improved, and the high accuracy and recall rate of video infringementretrieval are ensured.

The invention discloses a symmetrical ternary string represented voice perception Hash sequence constructing and authenticating method. The method comprises the steps that firstly, overall discrete wavelet transforming (DWT) is carried out on voice signals produced after preprocessing and intensity-loudness transformation (ILT); secondly, non-overlapping partitioning is carried out on the low-frequency part of the voice signals produced after DWT, and short-time logarithm energy of blocks is calculated to obtain the signal frequency-domain features; lastly, a final ternary perception Hash sequence is generated based on the time domain spectrum flux features (SFF) of the voice signals, and the voice frequency content is quickly authenticated through Hash matching. The symmetrical ternary string representation of the perception Hash abstract is superior to that of the binary form, the common voice content is operated between the robustness and the difference in a balanced mode, the time complexity of the algorithm is low, efficiency and the abstraction are high, precise manipulation detecting and positioning can be achieved, and the method can be used for authenticating a mobile voice communication terminal with bandwidth resources limited in real time.

A feature extraction and authentication method for multi-format audio perceptual Hashing authentication mainly aims to solve the problems that the audio format authenticated by an existing audio authentication algorithm is single, the authentication algorithm is not universal and authentication efficiency is low. The method is based on a dual tree complex wavelet transform (DT-CWT) principle, and includes the steps that global DT-CWT is carried out on preprocessed audio signals and a real wavelet coefficient of the audio signals is extracted; second order difference and framing are carried out on the real wavelet coefficient, and the sum of frame coefficients under a natural logarithm is calculated to serve as a feature vector of the audio signals; Hashing construction is carried out on the feature vector to generate a perceptual Hashing abstract sequence, and quick authentication of audio content is achieved through Hash match.

The invention relates to the technical field of medical image processing, in particular to a method for automatic freezing of a digestive endoscopy image based on a perceptual hash algorithm. The method comprises the following steps: S1, analyzing an endoscopy video stream acquired by endoscopy equipment into image data; S2, calculating the similarity between a picture at a time point t and previous n frames of pictures to obtain the weighted similarity k of the picture; and S3, comparing the picture weighting similarity k at the time point t with a freezing boundary l, and triggering an imagefreezing instruction when the value k reaches the value l. After the method is applied, when an endoscopic doctor needs to carefully check a view image, the endoscopic doctor only needs to stop moving an endoscope body to keep the view unchanged, a frozen image can be automatically judged, and the doctor does not need to manually operate a freezing button, so the workload of the doctor is reduced. A system automatically executes a freezing instruction, so offset of a frozen image view or loss of effective information caused by slow response or unskilled operation of people can be avoided, anda clear image of the optimal view can be effectively obtained.