64 results about "Fastica algorithm" patented technology

The invention provides a method for rapidly and automatically identifying and removing ocular artifacts in an electroencephalogram signal and belongs to the technical field of biological information and the method is mainly applied to a process of acquiring and preprocessing the electroencephalogram signal. The method comprises the following specific steps of: carrying out discrete wavelet transformation on an acquired multi-channel electroencephalogram signal and an electro-oculogram signal to obtain multi-scale wavelet coefficients; using the wavelet coefficients connected in series as an input for analyzing an independent component, and rapidly acquiring the independent component by using a negative entropy criterion-based Fast ICA (Independent Component Analysis) algorithm; identifying the ocular artifacts through a cosine method, performing zero resetting on the independent component, and projecting the other components through ICA inverse transformation and returning to all electrodes of an original signal; and finally obtaining the electroencephalogram signal for removing the ocular artifacts through inversion of the wavelet transformation. By utilizing the method for rapidly and automatically identifying and removing the ocular artifacts in the electroencephalogram signal, the problems that an ICA method is poor in discrete effect and low in convergence rate when beingapplied to noisy electroencephalogram signals are solved, and the function of rapidly and automatically identifying and removing the ocular artifacts in the electroencephalogram signal is realized.

The invention relates to a characteristic extracting method for prediction of rotating mechanical failure trend. The method includes the steps: (1) utilizing the remote online monitoring diagnostic center to conduct industrial onsite data collection and collecting vibration signals xj (t) of a plurality of channels through a plurality of sensors arranged on a rotating mechanical device; (2) conducting blind source separation on the vibration signals xj (t) according to FastICA algorithm and obtaining similar signal source yj (t) of the original independent vibration source sj (t); and (3) conducting characteristic frequency band decomposition of time frequency domain based on small wavelet packet on vector signals Y of the similar signal source yj (t) and extracting fault sensitive characteristic band. The characteristic extracting method is capable of recognizing the original independent signal source which shows as collecting signals in aliasing mode by adopting independent component analysis (ICA) processing, conducts characteristic frequency band acquisition based on the small wavelet packet on the independent signal source to judge whether one source signal has the development trend to fault and achieve the aim of preventing the fault in advance. The characteristic extracting method can be widely applied to prediction of the rotating mechanical failure trend.

The invention discloses a TFT-LCD mura defect detection method based on ICA learning and multichannel fusion, belonging to the TFT-LCD display defect detection field. The TFT-LCD mura defect detection method based on ICA learning and multichannel fusion comprises steps of utilizing an FastICA algorithm to separate mutually independent image bases from massive sample images, using the image bases to construct a background image, maintaining background information as much as possible without being influenced by the object, performing thresholding on the difference image of a test image and a background difference image, reducing the interference on the object partitioning which is caused by a potential object area and noise through setting a plurality of threshold values, introducing a detection scheme of multi-color channel fusion while considering the color information of the mura defect and giving consideration to mura defect detection of different types. The TFT-LCD mura defect detection method based on ICA learning and multichannel fusion is applicable to the mura defect detection of various types and reduces the over-detection and lead detection.

The invention relates to a novel ICA (independent component analysis)-based EMD (empirical mode decomposition) improvement process IMF (intrinsic mode function) judgment method. Aiming at the problem of excessive IMF components generated in the frequency band filtering EMD improvement process, the method introduces ICA into the EMD improvement process to automatically separate real IMF components. The method includes: improving EMD to decompose a structural response signal to obtain IMFs of each frequency band; and respectively taking the IMFs as an input matrix, and using a FastICA algorithm in the ICA for separation to automatically separate the real IMF out. The method can well process multi-degree-of-freedom, non-linear and unsteady-state response signals, can be combined with other methods (such as Hilbert transform) for modal parameter recognition, can be used for signal processing and modal parameter recognition in the fields of civil engineering, aerospace, automatic control, mechanical engineering and the like, and has the advantages of increase of signal to noise ratio of data, anti-jamming capability and the like.

The invention relates to a practical method for removing artifacts from online electroencephalograph, and belongs to the technical field of biomedical information processing. The method corrects a down sampling of the reduced channel real-time electroencephalograph signals, a power frequency notch wave and a linear drift, the discrete wavelet transform is used for decomposing the down sampling of the reduced channel electroencephalograph signals into 7 layers, and single channel electroencephalograph signals are converted to multiple channels. The wavelet coefficients are reconstructed and used as inputs to ICA. Fast acquisition of independent components is implemented using Fast ICA algorithm. According to the characteristics of time domain, frequency domain and ordinal correlation of each artifact in the independent component, which is different from the normal electroencephalograph component, hierarchical clustering algorithm is introduced to cluster each independent component, the categories of artifacts are automatically recognized, the artifacts are reconstructed after the zero artifact is zero, and the reconstructed electroencephalograph signals are obtained. The method solves the problem that the prior method cannot automatically identify and remove a variety of conventional electroencephalograph artifacts in the absence of reduced channel.

The invention discloses a voice detection method of power equipment failure based on a combined similar diagonalizable blind source separation algorithm. The method comprises the specific steps as follows: (1) adopting a microphone array; (2) separating all independent sound source signals from sound signals collected by the microphone array by adopting the combined similar diagonalizable blind source separation algorithm; (3) extracting Mel-MFC (Frequency Cepstral Coefficients) of the independent sound source signals as sound characteristic parameters, and identifying the sound signals through a model matching algorithm, wherein a reference sample template with a minimal matching distance is a result of identifying the operating sound of the power equipment after a sound template to be tested and all reference sample templates are matched. According to the voice detection method provided by the invention, the characteristics of a non-Gaussian source signal can be enhanced, a source signal which is more clear than a Fast ICA (Independent Component Analysis) can be estimated, the similarity coefficients of the separated signal and the source signals are 0.9 or above, the voice frequency audiometry on the separated signals can be carried out, and the signals separated by a JADE algorithm is clear and distinguishable.

The invention relates to a rolling bearing fault feature extraction method based on CEEMD and FastICA and belongs to the technical field of fault diagnosis and signal processing and analysis. The method comprises the following steps that: vibration signals are decomposed into IMF components with different frequencies through the CEEMD algorithm, corresponding IMF components are selected accordingto kurtosis criteria so as to be reconstructed into observation signals, and the residual IMF components are reconstructed into virtual noise channel signals; unmixing and denoising processing is performed on the observation signals and the virtual noise channel signals through the FastICA algorithm; demodulation processing is performed on the denoised signals through the Teager energy operator; and FFT (fast Fourier transformation) is performed on the demodulated signals, the frequency spectrum characteristics of the transformed signals are analyzed, the fault characteristic frequencies of the signals are extracted, and a fault diagnosis result is obtained. With the method adopted, the problem of fault information loss during a denoising process and the problem that noises cannot be completely removed due to modal aliasing can be solved; fault fundamental frequencies and frequency multiplication information can be extracted clearly and accurately; and the fault diagnosis result can beobtained.

A method for extracting the fault feature of an antifriction bearing based on a sliding entropy-ICA algorithm is provided. The method comprises main steps of: (1) subjecting a single-channel actually measured signal to EMD to obtain respective IMF components; (2) screening out effective IMF components by using a sliding entropy cross correlation coefficient to form a virtual channel signal; (3) integrating the single-channel actually measured signal with the effective IMF components to form a composite signal matrix, separating the composite signal matrix by using a FastICA algorithm to obtain respective source signal estimated values; (4) retaining a source signal containing a bearing fault feature, and extracting a plurality of time-domain feature parameters and frequency-domain feature parameters to form a feature parameter set; and (5) subjecting a high-dimensional feature set to data fusion by using an LLE algorithm to obtain an accurate low-dimensional feature parameter. The method uses the sliding entropy-ICA algorithm in combination with the LLE algorithm, is suitable for extracting the fault feature of rotating machines including the antifriction bearing, and can extract the source signal containing fault information just by using the single-channel signal. The low-dimensional feature parameter obtained by the method can describe bearing fault information.

The invention discloses a method for detecting and identifying raindrops in a video image. The method comprises the following steps of: firstly, detecting the raindrops by a FastICA algorithm according to the characteristic of raindrop motion; secondly, performing image binarization by a marginal information guidance-based process; and finally, identifying the raindrops according to the statisticcharacteristics of areas, direction angles and widths of a communication area in a binary image. The method can realize the detection and identification of the raindrops in the video image, is suitable for scenes of heavy rain, light rain, a dynamic state and a static state, and establishes the foundation for the removal of the raindrops in the video image.

The invention relates to an image quality evaluation method based on independent component analysis. The image quality evaluation method is characterized by being suitable for image quality evaluation of a grey scale map and a color image synchronously. The image quality evaluation method comprises the following steps of firstly, centrally training a group of ICA (Independent Component Analysis) decomposing matrixes from a reference image by utilizing a FastICA (Fast Independent Component Analysis) algorithm; secondly, multiplying each image block in the reference image and an image to be evaluated, and the ICA decomposing matrixes so as to obtain the independent component of each image block; lastly, measuring the quality of the image to be evaluated according to the difference of the independent components of the reference image and the image to be evaluated. In comparison with the conventional method, the method is capable of simulating expression of a visual signal in a human visual cortex and is closer to subjective image quality evaluation. The main calculated quantity of the method is centralized to the independent components, which are obtained by multiplying each split image block and the ICA decomposing matrixes, of the image blocks, but the calculation of each image block is independent, so that parallel computing is adopted, thus the execution efficiency is improved.

The invention discloses a water chilling unit fault detection method based on an improved FastICA. The water chilling unit fault detection method introduces relaxation factors on the basis of the traditional FastICA, changes the original iterative manner, and effectively reduces the dependence of the algorithm on initial values. The water chilling unit fault detection method comprises two stages of ''offline modeling'' and ''online detection''. The ''offline modeling'' stage comprises the steps of: firstly, carrying out steady state treatment on a normal data sample; then extracting independent component information of the data by adopting the FastICA algorithm which introduces the relaxation factors; and finally calculating statistics I2 and SPE, and determining control limits by adopting a kernel density estimation method. The ''online detection'' stage comprises the steps of processing newly acquired refrigeration process data according to a model, calculating statistics and comparing the statistics with the control limits to judge whether the refrigeration process operates normally. The water chilling unit fault detection method based on the improved FastICA reduces the dependence of the FastICA algorithm on initial values, and ensures the safety and stability of the water chilling unit in the refrigeration process.

The invention relates to a noise reduction method for a vibration signal of mechanical equipment. The method comprises the steps of: performing local mean decomposition on a non-stable vibration signal; according to PF components obtained by local mean decomposition, calculating a cross correlation coefficient of each PF component and the non-stable vibration signal, comparing the cross correlation coefficient with a preset value, and performing superposition reconstruction on each PF component when the cross correlation coefficient is smaller than the preset value to obtain a virtual noise channel signal, wherein the virtual noise channel signal serves as an input signal of an FastICA algorithm; and performing blind source separation on the vibration signal and the virtual noise channel signal according to the FastICA algorithm to obtain a source signal and a noise signal of the vibration signal, so that the noise reduction of the vibration signal is realized. The method can effectively reduce noise interference in the vibration signal, enable fault character frequency to be more obvious and then facilitate extraction of fault characteristics, and can be widely applied to the field of fault diagnosis of mechanical equipment.

The invention provides a fault diagnosis method for a fire pump operating at a low speed. The fault diagnosis method aims to achieve fault diagnosis of the fire pump operating at a low speed at the non-fire state. The fault diagnosis method includes the specific steps that a high-frequency stress wave sensor carries out signal detection on the fire pump at the inspection state; a high-frequency stress wave composite signal, containing noise, of the fire pump operating at a low speed is collected; the signal is processed through a FastICA algorithm, and a source signal is estimated; the Hilbert-Huang Transform is carried out on the estimated signal, and an n-order intrinsic mode function and a Hilbert spectrum of a fault signal are obtained to serve as a feature vector for judging the state of the fire pump and recognizing faults; finally, a support vector machine is utilized for classifying the faults. The fault diagnosis method is suitable for processing non-linear and non-stable signals under the situation of a low speed, more complete fault features can be obtained, and the fault diagnosis effect is better.

The present invention relates to a vibration signal combined denoising method. The method comprises the following steps of decomposing a wavelet transform modulus maximum of a vibration signal, and figuring out the positions of modulus maximum points and modulus maximum points corresponding to a wavelet transform coefficient on each scale; selecting a maximum wavelet decomposition scale J, adopting a pre-set threshold value T on the above scale as a search threshold, reserving the points of the modulus maximum points larger than T, removing the points of the modulus maximum points smaller than T and obtaining a new modulus maximum point on the scale J; searching a new modulus maximum point on the scale J-1 and positioned within a point set sequence in the neighborhood of the scale J; on the condition of the scale J=1, reserving the corresponding extreme value points of the scale J=1 at the positions of extreme value points of the scale J=2, while setting the values of extreme value points at all the rest positions to be 0; through the alternating projection process, obtaining a modulus maximum value on each scale and a reconstructed wavelet coefficient at the above position, and obtaining a reconstructed signal through inverse transform based on the obtained wavelet coefficient; based on the FastICA algorithm, subjecting a denoised signal and the vibration signal to blind source separation; and completing the combined denoising treatment on the vibration signal.

The present invention discloses a harmonic current estimation method based on an improved FastICA algorithm. The harmonic current estimation method is characterized in that a quasi-Newton method is adopted to replace a Newton method as an optimization method of FastICA. The present invention has the beneficial effects that (1) harmonic currents in a power grid can be accurately estimated in the condition of unknown harmonic impedance; (2) the problems that a harmonic measuring device is difficultly obtained by the actual harmonic impedance of a system, and the cost is great are overcome; and (3) compared to the original method, the harmonic current estimation method is capable of improving the accuracy of the algorithm and better solving the problem that the harmonic impedance is difficult to determine.

The invention relates to a method for eliminating an EEG signal noise artifact. The method comprises the steps of performing independent component separation on an EEG signal after denoising processing, then selecting an independent component with an artifact, and finally reconstructing the independent component which does not comprise the artifact for obtaining the EEG signal after noise artifactelimination, wherein independent component separation is based on an improved five-order FastICA algorithm. The improvement of the method is characterized in that an improved updating formula is usedfor updating the separated vectors and the relaxation factor of the independent component; and the independent component which comprises the artifact is selected based on a multi-domain adaptive threshold signal selecting method. According to the method for eliminating the EEG signal noise artifact, not only can a plurality of ocular artifacts be automatically identified and eliminated, but alsoa large amount of EEG information can be reserved. Furthermore the method can well improve the signal-to-noise ratio of the EEG signal and reduce the mean square error of the signal.

The invention discloses a fault diagnosis method for a switching current circuit based on improved FastICA. The fault diagnosis method comprises the steps that firstly, collected circuit frequency domain response data are subjected to centralization and whitening preprocessing; then an independent component analysis technology is utilized to find the maximum non-Gaussian direction of the data, animproved FastICA algorithm is applied to extract fault features of the switching current circuit, and a projection tracing technology is combined to achieve successful dimension reduction of originaldata; and finally, aiming at the two situations of not considering the circuit tolerance and considering the circuit tolerance, different diagnostic steps are given to complete fault classification and identification. Simulation experimental results show that a proposed fault extraction method can obtain the optimal fault features, and shows better performance than other feature extraction methods, and the fault diagnosis efficiency of the switching current circuit is advantageously improved.

The invention discloses a partial discharge ultrahigh-frequency signal blind source separation and denoising method based on principal component analysis. The method comprises the following implementation steps of: performing partial discharge detection on gas insulated switchgear by using an ultrahigh-frequency detection method to obtain original partial discharge ultrahigh-frequency signals; performing ensemble empirical mode decomposition on single-channel ultrahigh frequency signals to obtain limited intrinsic mode function components; performing spatial transformation on a matrix formed by the intrinsic mode function components by utilizing principal component analysis to obtain eigenvalues of the intrinsic mode function components, and arranging the eigenvalues from large to small; determining the number of source signals and constructing multi-channel detection signals in a new feature space by analyzing the variation trend of the eigenvalues; and carrying out blind source separation by using a FastICA algorithm based on independent component analysis, and obtaining denoised ultrahigh frequency signals. According to the method, environmental white noise and periodic communication noise can be effectively removed, the denoised signals are closer to noise-free source signals, the calculation amount is small, and the diagnosis accuracy is improved.

The invention relates to a measurement method for phase difference among same-frequency signals based on SOBI (Second Order Blind Identification) and FastICA (fast independent component analysis). The method comprises the following steps: 1. extending a tested signal x(n) into a three-dimensional 3D observation signal matrix X(n); 2. operating the sequenced non-iterative SOBI algorithm for the observation signal matrix X(n) once so as to obtain a separation matrix W1; 3. adopting W1 as an initial value of the separation matrix and operating the FastICA algorithm for the observation signal matrix X(n) once so as to obtain a hybrid matrix A and a source component matrix S(n); 4. synthesizing tested signal content x'(n) according to the hybrid matrix A and the source component matrix S(n); and 5. comparing the tested signal x'(n) with a standard signal so as to obtain a phase difference, thus completing the measurement of the phase difference among the same-frequency signals. The measurement method for the phase difference among the same-frequency signals based on the SOBI and FastICA provided by the invention can lower the requirements of measurement samples and increase the measurement accuracy.

The invention discloses a resonant sparse decomposition and FastICA algorithm-based planetary gear box fault diagnosis method. The resonant sparse decomposition and FastICA algorithm-based planetary gear box fault diagnosis method comprises the following steps of 1) decomposing a vibration signal to a high-resonant component and a low-resonant component by a resonant sparse decomposition method, and removing the low-resonant component containing a broadband signal; 2) taking the high-resonant component containing a planetary gear box as an observation signal, and performing resonant sparse decomposition on the observation signal to form a virtual channel signal; and 3) processing the observation signal and the virtual channel signal by a rapid independent component analysis algorithm, andseparating out an effective fault characteristic component so as to identify a fault type. By the resonant sparse decomposition and FastICA algorithm-based planetary gear box fault diagnosis method, the fault characteristic frequency of the planetary gear box can be effectively extracted, the problems of fault information loss and mode mixing during the EMD denoising process are solved, meanwhile,the problem of decomposition inaccuracy caused by difference between source signal numbers and observation signal numbers in ICA can be solved, and the fault characteristic frequency of the planetarygear box also can be accurately and clearly extracted.

The invention discloses a single-channel conduction electromagnetic interference noise separation method, using continuous wavelet transform and FastICA blind source separation EMI noise difference common-mode separation strategy, the implementation steps include: A, using CWT for L-line or N-line observation signals Implement virtual channel expansion; B. Use the improved FastICA algorithm to implement blind source separation for the extended signal to obtain differential common-mode noise separation signals; C. Evaluate the separation ability of differential common-mode noise signals and perform amplitude correction. Although the separation performance of the method is lower than that of the dual channel, it only uses half of the observation information, overcomes the dual channel synchronous measurement error, and saves economic costs, and has potential application value in the field of industrial EMI noise difference common mode separation.

An initialization method for image independent component analysis, it is: processes wavelet sparse decomposition for the receiving mixed signal; selects the decomposed coefficient group with the best sparseness, seeks the gather-axis to estimate the mixed matrix by the clustering method in the star-chart; processes initialization to the FastICA arithmetic by the estimating value of the mixed matrix, in order to avoid the minimum partly when the independent component analysis method, expedites the arithmetic constringency, simultaneously improves the separation precision of the independent component analysis method, gets the perfect image separating effect. The invention is fit for the radio-communication system, the sonar and radar system, the audio and acoustics signal process of the military field and non- military field.

The invention discloses a blind source separation technology controlled focusing system based on an FASTICA algorithm. The system comprises a blind source separation module, a GCC-PHAT time delay estimation module, a geometric positioning module and a camera module which are connected in sequence, wherein the blind source separation module comprises an amplifying circuit, a synchronous data acquisition circuit and a signal processing module; the GCC-PHAT time delay estimation module comprises a multichannel audio input circuit an a digital signal processor; the geometric positioning module comprises multiple voice channels and a digital signal processor and a control logic CPLD; and the camera module comprises an image information analysis and focusing weight adjustment module and a digital camera module. The system adjusts the distance between the internal parts of the camera lens and the photosensitive component according to the shooting distance measured during shooting, thus the subject can be clearly imaged on the photosensitive component, and the best focusing point can be obtained through adjustment even under low light level and low contrast conditions to achieve focusing. The system combines the blind source separation technology with the camera, thus the camera is simple to operate, accurate to position, intelligent and short in focusing time.

The invention provides a blind source separation method for noise reduction of vibration signals of a solid rocket engine. The number of source signals in acquired mixed signals under the condition that the source signals are unknown is distinguished according to an AIC criterion, and independent vibration source feature information is acquired from the mixed signals through observation signals ofa sensor according to a FastICA algorithm obtained after a learning function is improved. The problem about noise reduction of the vibration source signals under the condition that environments and equipment are complex is solved, compared with filtering and other traditional methods, the method better conforms to the solid engine test condition, useful source signals are not filtered out, the manual analysis error possibility is eliminated, the beneficial effects of intelligence, universality, adaptability and the like are achieved, two or more mixed mechanical vibration signals can be wellseparated, the purpose of noise reduction is achieved, the method can be applied to multiple engine test occasions, better parameter indexes are provided for engine designers, and the engine ground test requirements are met.

The invention relates to a method for improving human face video heart rate detection by using an illumination equalization method. The method comprises the following steps: S1, acquiring a human facevideo image by using a visible light camera; S2, completing face detection and positioning by using a multi-task convolutional neural network; S3, selecting a region of interest of the face video; S4, extracting a scene illumination component by using a fast guided filtering algorithm, constructing an improved two-dimensional gamma function, and balancing the illumination component of the face video image; S5, separating an independent source signal from the mixed signal by using a FastICA algorithm; and S6, performing fast Fourier transform by using the independent source signal, and calculating a heart rate value. According to the method, the illumination component is extracted through the fast guided filtering algorithm, the brightness of the too bright and too dark areas of the face image is improved through the improved two-dimensional gamma function self-adaptive correction illumination non-uniformity light equalization method, the average error and standard deviation of heart rate measurement values are reduced, and the measurement precision is improved.

The invention belongs to the signal processing technology field, and especially relates to a high-efficiency fast fixed point processing method suitable for communication signal blind separation. Aimed to the defect that a present non-circle fast fixed point algorithm (nc-FastICA) cannot selects an optimal non-linear function adaptively, an optimal non-linear function is selected adaptively through analysis of characteristics of different communication signals, and a fast fixed point processing method used for communication signal blind separation is provided. The separation efficiency is high, performances are stable, the convergence speed is fast, the provided is is raised by about 8-14dB compared with a traditional nc-FastICA algorithm and an ACMN algorithm when the separation communication signals are separated, and the provided method can be widely used for communication signal blind separation.

The invention discloses a seismic signal processing method and corresponding processing system based on an improved FastICA algorithm. Through maximizing a negative entropy object function, a non-Gaussian maximum value required by observing signal linear projection is searched by using a fixed point iteration theory so as to fulfill an aim of separating signals; an iteration function is correctedby introducing correction factors improved by utilizing a gradient method, and the FastICA algorithm is optimized. According to the improved FastICA algorithm, the sensibility to an initial weight value is reduced, and the iteration times are reduced, so that the convergence speed of the algorithm is improved.

The invention discloses a non-contact heart rate detection method. Firstly, chromaticity characteristics S1 (t) and S2 (t) of signals collected by a camera are calculated; pretreatment and extractionare carried out on each path of signals through the chromaticity characteristics S1 (t) and S2 (t) to obtain pulse signals h (t), h1 (t) and h2 (t); further blind source separation is performed on theextracted pulse signals h1 (t) and h2 (t) by adopting a FastICA algorithm; kurtosis values of power spectra of the pulse signals h (t), h1 (t) and h2 (t) are used as judgment of a blind source separation result; in addition, when the kurtosis values of the power spectra of the signals are close, only the selection signals with the kurtosis values prone to mistake are used, and the signal-to-noiseratio is used as an index, so that signal selection can be completed; therefore, noise interference caused by illumination change and non-autonomous movement of a subject can be effectively processed, a signal with lower noise is obtained, and the anti-interference performance and the measurement accuracy of detection are improved.

The invention provides an image encryption and decryption method for utilizing the gray system theory to confirm ICA output. The method comprises the steps of adopting the symmetric encryption algorithm based on ICA to conduct encryption work; adopting the Fast ICA algorithm to solve figures at the encryption end. If the sequence of encryption output images is not confirmed, using main components to analyze main components of the images, using the component data of the main components as model building data of a gray model, conducting gray association analysis on the images, and conducting Deng-type gray correlation degree calculation so as to obtain the gray correlation degree of the images. Sequences with the largest gray correlation degree are corresponding clear images and decrypted images. Image characteristic extracted through the method are accurate, and characterized data volume is not influenced by image blocks and image patterns, identification data volume is small, the algorithm speed is fast, manual threshold setting is not needed for judgment, and the problem that the decrypted output image sequence is not confirmed is well solved.

The invention discloses a no-reference image quality evaluation method based on independent component analysis. The method comprises the following steps: selecting an image quality block representingimage quality by utilizing spatial correlation of pixel pairs; Using the selected optimal image quality block and adopting a Fast ICA algorithm to obtain detection characteristics of images in an LIVE2 image library, and obtaining independent components of the image quality block according to the characteristics; Establishing a hash lookup table by taking binaryzation of independent components ofan image in LIVE2 as a hash function, wherein each data element of the lookup table comprises judgment data, independent component data and a subjective quality evaluation (DMOS) value; And searchingthe hash table by using binaryzation of independent components of the image to be detected, and performing Hamming distance matching on the conflict item to obtain a no-reference image quality evaluation result. The evaluation result and the DMOS value keep good consistency, and the distortion degree of the image can be measured accurately.