63 results about "Quaternion matrix" patented technology

The invention relates to a method and system for implementing playback of a panoramic video on a mobile device. The method comprises the steps of after sensing that a user acquires a selected video image by a trigger operation on a video application, creating a three-dimensional sphere in a virtual environment by using an OpenGL ES (Open Graphics Library for Embedded System); placing a camera in the three-dimensional world inside the sphere; executing a video acquiring and converting step, i.e. reading the video image at a preset speed, acquiring each frame of video image and converting the video image into texture information; executing a texture information mapping step, i.e. mapping the texture information obtained by conversion into the sphere; executing a view angle calculating step, i.e. according to a rotating angle and a camera view angle of the three-dimensional sphere, carrying out quaternion matrix calculation to obtain an actually displayed image matrix; and repeating the video acquiring and converting step, the texture information mapping step and the view angle calculating step, and acquiring a panoramic video image. Therefore, 360-degree panoramic video playback on the mobile device is implemented, and a video viewing effect is improved.

The invention discloses a color image quality evaluation method based on HVSs and quaternions and belongs to the technical field of image processing and computer vision. The method includes the steps that firstly, mathematic evaluation models of an original reference image and a distorted image to be evaluated are established through analysis of human vision features, wherein the mathematic evaluation models comprise spatial location functions Q<L>, local variances Q<V>, texture edge complexity functions Q<TE> and color functions Q<C> of the images; secondly, quaternion arrays of the original reference image and the distorted image to be evaluated are established and singular value decomposition is conducted on the quaternion arrays, so that singular value feature vectors of the images can be acquired; thirdly, the image distortion degree is measured through the Euclidean distance of the singular value feature vectors of the original reference image and the distorted image to be evaluated. According to the method, the human vision features and the quaternions are combined, brightness and chromaticity information of the images is extracted, the spatial location functions, the texture edge complexity functions and the local variances are established through the human vision features, and an evaluation result accords with a result generated in the mode that the images are sensed by the human eyes.

The invention discloses a multi-focus image fusion method based on quaternion wavelet transform in order to more accurately integrate the information in a source image into a fusion image. The method comprises steps of: 1) subjecting two multi-focus images to be fused to quaternion wavelet transform in order to obtain a low-frequency sub-band (LL) and three high-frequency sub-bands (LH, HL, HH) of each layer, wherein each sub-band corresponds to four quaternion coefficient sub-bands, the four quaternion coefficient sub-bands can be transformed to an amplitude sub-band and three phase sub-bands by quaternion algebraic operation; 2) subjecting the low-frequency sub-band to fusion by a fusion rule based on improved phase gradient characteristic value maximization, and subjecting the high-frequency sub-bands to fusion by a fusion method based on quaternion matrix comprehensive multi-features; and 4) subjecting the fused high and low frequency sub-band coefficients to quaternion wavelet transform to obtain a fused image. The method can fully extract the features of the source image and accurately represent the source image so as to avoid introducing error information into the fused image and to improve a visual effect, and greatly improves the quality of the fused image compared with a traditional fusion method.

The invention discloses a method for determining and compensating a full-field drift angle of a space-based camera. The method comprises the following steps: (1) performing scene simulation and modeling, comprising performing high-precision modeling and simulation on the satellite orbit, attitude, camera measurement coordinate system, satellite body measurement coordinate system and different field sensors of the camera; (2) calculating the full-field drift angle of the camera under a geocentre terrestrial reference frame; (3) calculating a compensation angle of the satellite body measurement coordinate system; (4) calculating a compensation attitude quaternion matrix on the satellite drift angle; and (5) calculating the full-field drift angle of the camera subjected to attitude compensation. According to the method disclosed by the invention, after a three-axis stability satellite rotates around three axes of the satellite body, the problem that the full-field drift angle of the camera cannot be determined and compensated during earth observation of the satellite in a specific attitude can be solved, and the method can be used for determining the drift angle of a satellite-borne large-field camera, designing and verifying the drift angle compensation method and optimizing the satellite design in the satellite development process.

The invention discloses an objective video quality evaluation method based on space domain and time domain structural similarities. Structural similarity between a distorted video and a reference video is evaluated from two aspects of the space domain and the time domain, space domain evaluation comprises luminance and chrominance structural similarities, the time domain evaluation is presented by using interframe residual structural similarities and calculation is simple and quick. A quaternion matrix is then used for building a structural similarity evaluation model. A singular value decomposition method is then adopted for solving the structural similarity quaternion matrix and the final objective quality evaluation value is obtained. The method has the advantages that complexity of calculation is low, and correlation between the objective evaluation result and subjective perception can be effectively improved.

The invention discloses a quaternion field colored image compressed sensing recovery method based on a Quasi Newton algorithm. The quaternion field colored image compressed sensing recovery method comprises the following steps of: converting a colored image two-dimensional matrix to a two-dimensional matrix of the quaternion field; and carrying out compressed sensing on the two-dimensional matrix of the quaternion field once to recover the original colored image, so that the operation time is saved in comparison with a conventional method of carrying out compressed sensing on the three vectors, i.e. RGB (Red, Green, Blue) of the colored image respectively. According to the quaternion field colored image compressed sensing recovery method based on the Quasi Newton algorithm disclosed by the invention, a quaternion matrix signal is written into an Euler form of a quaternion, the amplitude and the phase are adopted as novel bound terms of compressed sensing optimization problems, therefore, the recovery effect is better in comparison with the conventional method which converts the data of the three channels, i.e. RGB of the colored image into three real-number two-dimensional matrixes for being processed respectively. According to the quaternion field colored image compressed sensing recovery method based on the Quasi Newton algorithm disclosed by the invention, the result image is a mean value of an image obtained by carrying out compressed sensing decoding in rows and an image obtained by carrying out compressed sensing decoding in columns, therefore, the obtained image is smoother than the image which is recovered by being processed only in rows or in columns.

The invention discloses an expression recognition method and an expression recognition system for privacy protection. The method comprises the following steps: obtaining a plurality of color face images, and carrying out the preprocessing of the plurality of color face images, and obtaining the color components of the plurality of color face images; taking the color components as imaginary parts of a quaternion matrix, constructing a first quaternion matrix, and calculating a local variance of the first quaternion matrix; constructing a second quaternion matrix by taking the local variance asa real part of the quaternion matrix and the color component as an imaginary part of the quaternion matrix; performing encryption processing on the second quaternion matrix by adopting quaternion Gyrator transformation to generate a real number matrix; randomly scrambling the real number matrix so as to train the convolutional neural network, and using the trained convolutional neural network to identify and verify the color face image. The method has the advantages of stronger recognition performance, higher safety and revocation.

The invention provides a textural feature extraction method. The textural feature extraction method includes the steps that an input image is obtained and converted into an HIS color space; a simple double-quaternion matrix of the image converted into the HIS color space is calculated; a preset multichannel filter functional matrix is obtained; a convolution matrix obtained by calculating the convolution of the simple double-quaternion matrix and simple double quaternions of the multichannel filter functional matrix; pixel coordinates with adjacent difference values smaller than a difference value threshold value in the convolution matrix are searched for, and textural features are generated according to the searched pixel coordinates. In addition, the invention provides a textural feature extraction device. The textural feature extraction method and device can reduce the false drop rate of textural feature extraction.

The invention discloses a method for evaluating the quality of a full reference color image based on quaternion. The method includes the following steps: constructing a brightness channel component template of any pixel point; respectively calculating the local variance of an original reference color image and a to-be-evaluated distorted color image; obtaining a local variance V, a brightness channel L, a first color channel A and a second color channel B; taking V as a real portion of a quaternion matrix, and L, A, and B as virtual portions of the quaternion matrix, respectively constructing the quaternion matrix of the original reference color image and the to-be-evaluated distorted color image; conducting singular value decomposition on the quaternion matrices to obtain corresponding singular value characteristic vector; and calculating the Euler's distance of the singular value characteristic vectors, and taking the Euler's distance as the distortion level metric of the to-be-evaluated distorted color image. According to the invention, the method can better evaluate the distortion of the color image, requires less calculation, and can effectively increase the relativity between a subjective evaluation result and objective perception, and produces the result that has the precision as human eyes.

The invention discloses an algorithm for evaluating color image quality. The algorithm comprises: firstly performing HSI transformation on a reference image and images to be evaluated; representing the images with hue, intensity, and saturation; using the hue, the intensity, and the saturation as the imaginary part of a quaternion; constructing and performing singular value decomposition on quaternion matrixes of the reference image and the images to be evaluated in order to obtain singular value characteristic vectors; finally using grey relational degree to calculate relational degree between the singular value characteristic vector of the reference image and the singular value characteristic vector of each image to be evaluated, wherein the more the relational degree is, the better the quality of the image to be evaluated is. The validity of the algorithm is verified by actual examples and an evaluation result of the algorithm is better than conventional methods of MSE, PSNR, and SSIM.

The invention discloses a stereo image vision significant extraction method. A left viewpoint image of a test stereo image is converted to the Lab color space, and three channel images and a normalization image of a left parallax image are zoomed to 200*200 pixel dimensions; respective gradient amplitude graphs of the four images are acquired, and logarithm calculation is further carried out; center surrounding reinforcement of the images after logarithm calculation is carried out through utilizing a center preference map, the images after center surrounding reinforcement form a quaternion matrix, and quaternion Fourier transform of the quaternion matrix is carried out to acquire a frequency domain matrix; a frequency domain filtering template graph is then utilized to carry out low pass filtering of the frequency domain matrix, and quaternion Fourier inverse transform of the acquired low pass characteristic graph is carried out; lastly, a preliminary vision significant graph is acquired according to the quaternion matrix acquired through inverse transformation, and the preliminary vision significant graph is normalized and zoomed to W*H pixel dimensions to acquire a final vision significant graph. The method is advantaged in that relatively strong extraction stability and relatively high extraction accuracy are realized.

A partial feedback filter method for an integrated navigation system involves a GPS receiver, an inertial navigation system and an integrated filter. The method comprises the following steps: treatingthe speed and location information error of the GPS receiver and the inertial navigation system as kalman filter estimation value; multiplying kalman filter estimation value by a number factor a through an integrated filter; then feeding back to the inertial navigation system for parameter correction; converting a primary attitude matrix and a quaternion through the inertial navigation system toobtain a primary quaternion matrix; outputting and correcting the primary quaternion matrix of the inertial navigation system through the integrated filter; and calculating the output result of the integrated navigation system. With the adoption of the method, the location output accuracy of the navigation system is improved; the kalman filter variance matrix is limited; the output result is out of distortion; and the adaptability is improved.

The invention provides a biological feature recognition method and device, and relates to the technical field of image processing and recognition. The method comprises the steps of obtaining a to-be-detected image; Projecting the first full quaternion matrix representing the to-be-detected image on a preset feature vector to obtain a projection matrix; Wherein the preset feature vector is a feature vector of a scrambling matrix of a second full quaternion matrix of the training image set; And using the obtained projection matrix to classify and identify the obtained to-be-detected image, and obtaining an identification result. According to the method, the image information amount borne by the quaternion matrix is increased, and the face color image can be expressed more accurately, so thatthe recognition precision is further improved; The realization process of the scrambling matrix is irreversible, and even if the registered or stored biological characteristic data is lost or stolen,only the biological characteristic data needs to be registered or stored again, so that the security and privacy of the biological characteristic data such as a color face image are effectively protected.

The invention discloses a quaternion discrete cosine transform-based color image quality evaluation method, and mainly solves the problem of low consistency of objective evaluation and subjective evaluation of color image quality at present. According to the scheme, the method comprises the steps of 1) representing color three-channel information of a color image by utilizing a quaternion matrix;2) obtaining a coefficient of the image by utilizing quaternion discrete cosine transform; 3) according to a spatial frequency position, dividing the obtained quaternion color spectrum coefficient into different sub-bands; and 4) calculating the similarity between the sub-bands of an original image and a distorted image, and obtaining image quality through weighted fusion. The calculation is simple; an experimental result shows that three channels of the color image are subjected to overall processing by utilizing a quaternion; the color image can be subjected to quality evaluation more effectively and accurately; and the method can be used for processing the color image in compression, storage and transmission of the color image.

The invention relates to provides an image feature extraction method and device, and relates to the technical field of expression identification. The method comprises that a first face expression image is represented into a quaternion matrix; three phase matrixes and an amplitude matrix of the quaternion matrix are obtained; the three phase matrixes are coded partially to obtain three coded phasematrixes respectively, and the amplitude matrix is coded partially to obtain a coded amplitude matrix; and element values of each of the three coded phase matrixes and the coded amplitude matrix are counted to obtain an image feature of the first face expression image. According to the method, color information, grayscale information and internal connection therebetween included by the face expression image are taken into full consideration when the image feature is extracted, the extracted image feature can represent the expression feature in the image effectively, and a better identificationeffect can be achieved when the image feature is used for expression identification.

The invention discloses a method for recovering a quaternion field color image in a low sampling frequency, and belongs to the technical field of digital picture processing. Firstly, a color image is expressed in a quaternion field, namely, values of an R component, a G component and a B component of the color image are respectively placed in three imaginary parts of a quaternion; and then downsampling is conducted to an original whole color image by utilizing an even stochastic matrix, and part of the sampled color image is obtained; furthermore, part of the color image is stored or conveyed; and finally the original color image is recovered through the part of the sampled color image by utilizing a real number expression form of the quaternion and quaternion matrix filling theory. The method for recovering the quaternion field color image in the low sampling frequency improves precision of recovering of the color image in the low sampling frequency and meanwhile simplifies complexity of sampling and storage content of data by utilizing a characteristic that a quaternion algorithm is good in binding force.

The invention discloses a high dynamic color image quality evaluation method based on feature similarity, The method comprises the following steps: 1, converting the original brightness into perceptual brightness, 2, calculating PC_R, PC_G and PC_B R of R of G and B color components of the color LDR image identical in phase, 3, calculating the gradient magnitude (GM) of R, G and B components of color LDR images GM_R, GM_G and GM_B; 4, integrating the PC feature map and the GM feature map into a quaternion matrix, and converting the feature map of color LDR images into pure imaginary form, 5, calculating feature similarity and realize quality evaluation. The invention retains the correlation of R, G and B color channels, and has high consistency with the subjective quality evaluation of human eyes. The algorithm has high evaluation accuracy for different distortions and low computational complexity, so it can achieve real-time quality evaluation.

The invention discloses an image fuzziness assessment method based on a quaternary phase congruency model. The method is represented by a color image based on a quaternion matrix, the color image is processed as a whole in the form of a vector signal and a quaternary phase congruency picture is calculated, and the quaternary phase congruency picture is combined with a traditional complex number phase congruency picture to form a significance picture which can be used to describe the structure contour clear degree of the image, that is a mixed phase congruency picture. According to the invention, histogram analysis is performed on the calculated significance pictures at the same time, the weighted average of phase congruency is taken as the index for measuring the image clear degree, and finally, the fuzzy coefficient of the image is calculated. The method of the invention is a significance method capable of reflecting the contour structure information of the color image, by effectively utilizing the correlation between color channels and the human visual characteristics, the color image fuzziness assessment accuracy rate can be improved.

The invention discloses an objective video quality assessment method based on chromaticity co-occurrence matrix weighting. The method includes the steps that firstly, a quaternion matrix with the dimension being 8*8 is used for describing characteristics, based on the time domain and the space domain, of 8*8 image blocks in a reference image and characteristics, based on the time domain and the space domain, of 8*8 image blocks in a distortion image; then the cosine of an included angle between singular value vectors of the characteristics, based on the time domain and the space domain, of two image blocks, which correspond to each other in position, in the reference image and the distortion image is obtained, and the cosine serves as the difference degree between the two image blocks, which correspond to each other in position, in the reference image and the distortion image; objective assessment values of the image blocks in the distortion image are obtained through the difference degree and weighted values of chromaticity co-occurrence matrixes of the image blocks in the distortion image, and therefore the objective assessment value of the distortion image and the objective assessment value of distortion video are obtained. According to the method, computation complexity in the video assessment process is low, and the relevance between an objective assessment result and subjective perception can be effectively improved.

The invention discloses a stereoscopic-image visual significance extraction method based on frequency domain sparse representation. The method is characterized by acquiring the small-sized three-channel image of the left-viewpoint image of a testing stereoscopic image and a small-sized normalized parallax image; carrying out partitioning on the three-channel image and the normalized parallax imageand acquiring matrixes corresponding to partitioning images; using the four matrixes to form a quaternion matrix, acquiring a transformed quaternion matrix after two-dimensional quaternion Fourier transform, and according to the low frequency components of a real portion and three imaginary portions, acquiring a low frequency component image; using a sparse expression dictionary to extract the sparse weight matrix of the low frequency component image; according to the sparse weight matrix and a center preference image, acquiring two center and periphery significance images; carrying out fuzzyprocessing on the two center and periphery significance images and then fusing and acquiring a fusion image; using the center preference image to reinforce the center and periphery of the fusion image; and acquiring a visual significance image after center and periphery reinforcement image size conversion. The method has advantages of high extraction stability and good extraction accuracy.

The invention provides an image Hash algorithm based on quaternion matrix singular value decomposition, which comprises the following steps: S1, reading a color image, and zooming the size of the color image; s2, representing the scaled color image by using a quaternion matrix to obtain a complex matrix; s3, equally dividing the complex matrix into (shown in the description) sub-blocks with the same size; s4, calculating the maximum singular value of each sub-block to obtain a numerical matrix with the size (shown in the description); s5, converting the matrix into a 01 matrix; and S6, performing calculating according to the 01 matrix to obtain the hash value of the color image. According to the algorithm, each channel in the image can be regarded as a whole to be processed, so that the correlation among the channels is brought into full play, and the color image processing capability is improved. In addition, the invention further provides an image similarity measurement method basedon the algorithm, the image similarity measurement method can solve the problem of color image similarity measurement, and the color distinguishing effect on color images is achieved.

The invention discloses a color image quality evaluation method based on HVSs and quaternions and belongs to the technical field of image processing and computer vision. The method includes the steps that firstly, mathematic evaluation models of an original reference image and a distorted image to be evaluated are established through analysis of human vision features, wherein the mathematic evaluation models comprise spatial location functions Q<L>, local variances Q<V>, texture edge complexity functions Q<TE> and color functions Q<C> of the images; secondly, quaternion arrays of the original reference image and the distorted image to be evaluated are established and singular value decomposition is conducted on the quaternion arrays, so that singular value feature vectors of the images can be acquired; thirdly, the image distortion degree is measured through the Euclidean distance of the singular value feature vectors of the original reference image and the distorted image to be evaluated. According to the method, the human vision features and the quaternions are combined, brightness and chromaticity information of the images is extracted, the spatial location functions, the texture edge complexity functions and the local variances are established through the human vision features, and an evaluation result accords with a result generated in the mode that the images are sensed by the human eyes.

The invention discloses a color face recognition method based on quaternion non-convex penalty sparse principal component analysis, and belongs to the field of mode recognition and artificial intelligence. The method comprises the following steps: firstly, encoding red, green and blue channels of a color image by utilizing quaternion, constructing a complex representation form of a quaternion matrix, then calculating a quaternion covariance matrix, and carrying out characteristic decomposition on the quaternion covariance matrix to obtain a principal component vector; and introducing a non-convex penalty L1 / 2 norm as a sparse constraint term to obtain a new quaternion non-convex penalty sparse principal component analysis (QHSPCA) optimization model, and solving a sparse solution of the model by adopting a coordinate descent method and an immobile point iteration method, and finally realizing face recognition by using a nearest neighbor classifier. Experiments on a Geoga Tech face database show that the QHSPCA method provided by the invention has relatively good recognition performance, and the calculation efficiency is also improved.

The invention provides a multichannel image encryption method and device, a storage medium and an electronic device. The method comprises the steps of encoding a to-be-encrypted multichannel image into a quaternion matrix; generating a chaotic phase mask according to the multi-channel image to be encrypted; performing quaternion transformation on the quaternion matrix in combination with a chaoticphase mask; determining a real value matrix according to the quaternion matrix after quaternion transformation; and determining the ciphertext based on the real value matrix. The chaotic phase mask is generated based on the multi-channel image to be encrypted, and quaternion transformation is carried out, so that the reliability of image encryption can be effectively ensured. The real value matrix covering the real component and the imaginary component of the converted quaternion matrix is determined to determine the ciphertext, so that the reliability of image encryption can be further improved, the multichannel components of the multichannel image can be encoded into a whole, the transmission efficiency can be effectively improved, and the storage is convenient.

The invention discloses an image encryption method and device based on cascade transformation, and the method comprises the steps: S1, generating a phase mask according to a chaotic sequence; s2, randomly generating a real integer matrix, and encoding the real integer matrix and the first color image to generate a full quaternion matrix; s3, performing modulation and quaternion Gyrator transformation on the full quaternion matrix according to the phase mask to obtain a first amplitude and a first phase; s4, encoding the first amplitude and the second color image to generate a quaternion matrix, and performing amplitude truncation to obtain a second amplitude and a second phase; s5, multiplying the first phase by the second phase, and performing modulation and quaternion Gyrator transformation on the quaternion matrix according to a phase mask to obtain a third amplitude and a third phase; s6, repeatedly executing S4 and S5 until the Nth amplitude and the Nth phase of the last color image are obtained. The method has relatively high transmission efficiency and security.

The invention discloses a method for detecting red tide and toxicity of seawater algae based on a three-dimensional fluorescence spectrum. The method comprises the following steps: irradiating to-be-detected seawater by monochromatic ultraviolet light; collecting seawater reflected fluorescence; enabling the reflected fluorescence to generate the three-dimensional fluorescence spectrum via a diffraction grating; converting the three-dimensional fluorescence spectrum in an optical signal form into a three-dimensional fluorescence spectrum in an electric signal form by virtue of a CCD (Charge Coupled Device) so as to obtain original three-dimensional fluorescence spectrum data of the seawater; performing parallel representation on a quaternion matrix of the original three-dimensional fluorescence spectrum data; extracting spectral features based on quaternion principal component analysis; and establishing a nonlinear mapping model for determining hemolytic activity of red-tide algae, substituting the extracted spectral features into the established model, outputting the hemolytic activity value by the model, and determining the toxicity of the algae based on the value. The method disclosed by the invention is used for detecting based on a spectroscopic method, and no chemical reagent is added; the equipment used by the detection method is small and high in integrated level, and continuous in-situ online measurement can be realized; and the detection accuracy is high.

The invention relates to the technical field of image processing, in particular to an image encryption method and device. The image encryption method comprises the steps of obtaining a quaternion matrix of an image to be encrypted; Determining a parallel component matrix and a vertical component matrix of the to-be-encrypted image based on the quaternion matrix; Determining a complex value mask ofthe to-be-encrypted image based on the parallel component matrix and the vertical component matrix; And generating an encrypted image corresponding to the to-be-encrypted image based on the complex value mask. By adopting the image encryption algorithm, the concentration of encrypted information can be avoided, so that the problem of obvious contour highlighting is further solved, and the security of the image in transmission can be improved.

The invention provides an image processing method, device and equipment and a storage medium. The method comprises the following steps: determining a quaternion matrix corresponding to each frame of image in multiple frames of images, wherein the multiple frames of images comprise a first frame of image and a second frame of image; processing the quaternion matrix corresponding to the first frameimage to obtain a first amplitude and a first phase of the first frame image; processing the quaternion matrix corresponding to the second frame image to obtain a second amplitude and a second phase corresponding to the second frame image; and taking the first amplitude and the second amplitude as a ciphertext of the multi-frame image, and generating a key of the multi-frame image according to thefirst phase and the second phase. According to the invention, the first amplitude, the first phase, the second amplitude and the second phase are quickly determined, and the secret key of the multi-frame image is generated according to the first phase and the second phase, so that the security of the encrypted image is higher, and the transmission efficiency is effectively improved.

The invention provides an expression recognition method and device and an expression recognition model training method and device, and the method comprises the steps: processing a to-be-recognized image, so as to obtain a face image; representing each pixel point in the face image by using a quaternion to form a face quaternion matrix, with the weight of each imaginary number in the quaternion representing the value of a corresponding color component; and taking the human face quaternary matrix as the input of a target neural network model, and recognizing the to-be-recognized image to obtainthe expression of the human face in the to-be-recognized image.