CN111107360B - Spectrum-space dimension combined hyperspectral image lossless compression method and system - Google Patents
Spectrum-space dimension combined hyperspectral image lossless compression method and system Download PDFInfo
- Publication number
- CN111107360B CN111107360B CN201911420225.1A CN201911420225A CN111107360B CN 111107360 B CN111107360 B CN 111107360B CN 201911420225 A CN201911420225 A CN 201911420225A CN 111107360 B CN111107360 B CN 111107360B
- Authority
- CN
- China
- Prior art keywords
- channel
- image
- hyperspectral
- uniformity
- transformation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/13—Adaptive entropy coding, e.g. adaptive variable length coding [AVLC] or context adaptive binary arithmetic coding [CABAC]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/12—Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264
- H04N19/122—Selection of transform size, e.g. 8x8 or 2x4x8 DCT; Selection of sub-band transforms of varying structure or type
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- Discrete Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
Abstract
The invention discloses a spectrum-space dimension combined hyperspectral image lossless compression method and a system, firstly, preprocessing an original hyperspectral image, and correcting nonlinear factors caused by an imaging instrument; reversible integer transformation is carried out on the preprocessed hyperspectral images, and correlation among spectrums is removed; calculating image uniformity indexes of the hyperspectral images subjected to reversible integer transformation channel by channel; the image channel with relatively uniform uniformity is coded by adopting a FILF compressor, and the image channel with poor uniformity is switched to an arithmetic coder for coding; decoding by a corresponding decoding mode, removing the inter-spectrum correlation, and performing reversible integer inverse transformation on the image without the inter-spectrum correlation to recover the original hyperspectral image before compression. The invention removes the relativity between spectrums by performing reversible integer transform on the hyperspectral image so as to reduce the information entropy of the original data and remove the redundant information between spectrums, adopts FILF compression coding to extract high compression ratio, and adopts a simple arithmetic coder to accelerate the compression speed.
Description
Technical Field
The invention belongs to the field of image processing, and particularly relates to a spectrum-space dimension combined hyperspectral image lossless compression method and system.
Background
The spectrum rate of the hyperspectral remote sensing reaches the nanometer magnitude, and the number of wave bands in the interval from visible light to short-wave infrared spectrum reaches dozens to hundreds. The high spectral resolution enables the hyperspectral image to provide more precise ground feature detail information, and the hyperspectral image can be widely applied to the fields of geological investigation, ore deposit detection, fine agricultural ocean remote sensing, environment and disaster monitoring, military reconnaissance and the like. In the development process of the hyperspectral remote sensing technology, along with the continuous improvement of spectral resolution and spatial resolution, the data volume acquired by an imaging spectrometer expands rapidly, and huge pressure is brought to the storage and transmission of data.
Because the information of the remote sensing image is precious, lossless compression is adopted as much as possible. Lossless compression does not allow any loss of the original image information, and there is no error between the image restored by decoding and the original image. The lossless compression is compression of the file itself, and like compression of other data files, the data storage mode of the file is optimized, a certain algorithm is adopted to represent repeated data information, the file can be completely restored, the content of the file cannot be influenced, and for image data, the details of the image cannot be lost. Lossless compression can therefore be considered as a reversible process.
In the field of still visible image compression, a unified international compression standard has been established, while in the field of hyperspectral image compression, a compression standard has not been formed. At present, popular visible light compression systems such as PNG, JPEG-LS, JPEG2000 and the like are mostly and directly applied to the compression of satellite-borne multispectral or hyperspectral images, the visible light image compression systems can only compress single-channel gray-scale images or three-channel RGB images, can only eliminate spatial correlation in image spectrums, do not consider correlation among wave bands, are low in compression performance, and cannot effectively reduce transmission bandwidth of data. Regarding how to eliminate the correlation among spectrums, the field of hyperspectral image compression has a plurality of prediction-based methods, a simple predictor (such as linear prediction) is often used, the prediction error is not reduced, the compression capability is weak, and a complex predictor (such as BP neural network) is used, although the smaller prediction error can be ensured, the model volume is larger, the volume of the model file is calculated, and the volume of the compressed image file is not reduced or increased. In addition, a large amount of compression time is wasted by simply and directly applying a popular visible light compressor to compress the hyperspectral image, so an effective and practical hyperspectral image data lossless compression method is urgently needed in the field.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a spectrum-space dimension combined hyperspectral image lossless compression method and system, and aims to solve the problem that the compression performance is limited because the FILF compression system in the prior visible light image field is still used in the hyperspectral image data lossless compression technology, so that the strong inter-spectrum correlation of hyperspectral images cannot be removed.
To achieve the above object, according to an aspect of the present invention, there is provided a spectrum-space dimension combined hyperspectral image lossless compression method, including the following steps:
(1) preprocessing an original hyperspectral image, and correcting nonlinear factors caused by imaging;
(2) reversible integer transformation is carried out on the preprocessed hyperspectral images, the correlation among spectrums is removed, and a transformation matrix is stored;
(3) calculating the image uniformity of the hyperspectral images subjected to reversible integer transformation channel by channel, and dividing the hyperspectral images into more uniform image channels and image channels with poor uniformity after determining uniformity evaluation indexes;
(4) the more uniform image channels are coded by a FILF compressor, the image channels with poor uniformity are coded by an arithmetic coder, and lossless compression of the hyperspectral image is realized;
(5) decoding the compressed hyperspectral images channel by channel, and decoding the images without the inter-spectral correlation by adopting an arithmetic encoder to encode and adopting an image channel to be decoded encoded by a FILF compressor in a corresponding decoding mode;
(6) and (3) reading the transformation matrix obtained in the step (2), and performing reversible integer inverse transformation on the image without the inter-spectrum correlation to restore the original hyperspectral image before compression.
Further, the step (2) specifically comprises:
(2.1) calculating a Pearson correlation matrix Coeff among the preprocessed hyperspectral image data spectrums:
where ρ (p)i，pj) The Pearson correlation coefficients of the ith and jth spectral bands are expressed, and the specific calculation formula is as follows:
wherein, Cov (p)i，pj) Denotes the covariance between spectral bands, D (p)i)、D(pj) Is the variance of the ith and jth spectral bands;
(2.2) carrying out principal component analysis on a Pearson correlation matrix Coeff to obtain a transformation matrix A, wherein the transformation matrix A can only realize the transformation from real numbers to real numbers, and the transformation matrix A obtained through decomposition is PLUS to realize the mapping from integers to integers, L and U are respectively a unit lower triangle matrix and a unit upper triangle matrix, P is a permutation matrix, and S is a unit element reversible matrix. The specific decomposition process of A is as follows:
let A be the nonsingular matrix that satisfies the decomposition condition:
(2-3) first, there is a row transformation matrix P1Make it
First element of the Nth column
Is not zero, s can be found1Make it
Namely, it is
Thus, a unit element invertible matrix S is constructed1：
Then there is
Then, a traditional Gaussian elimination method is adopted to define an elementary Gaussian matrix:
then there are:
(2-4) repeating the above process N-1 times to obtain:
LN-1PN-1…L2P2L1P1AS1S2…SN-1＝DRU
wherein k is 1, 2, … N-1; dRIs a twiddle factor, DR＝±1
(2-5) the transformation matrix A can be finally decomposed into the following formula:
A＝PLUS
wherein:
P＝(PN-1…P2P1)T
U＝±LN-1PN-1…L2P2L1P1AS1S2…SN-1
and (2-6) since L is a lower triangular matrix, U is an upper triangular matrix, and only the last row of the S matrix has valid data, the LUSs can be combined into a real matrix to be stored.
(2-7) sequentially multiplying the hyperspectral image data by the S, U, L and P transformation matrixes, and rounding each time the matrix multiplication is completed to realize integer KLT transformation, wherein the specific formula is as follows:
Y＝P*round(L*round(U*round(S*data)))
where round denotes the rounding operation.
Further, the step (3) specifically comprises:
(3.1) calculating and saving the minimum value of the hyperspectral images subjected to reversible integer transformation channel by channel, and subtracting the minimum value of the corresponding channel by channel so that the channel image data are positive integers which are more than or equal to 0;
(3.2) checking channel-by-channel whether the channel image data obtained after the processing of the step (3.1) has the maximum overflow problem, and if the channel image data has overflow, disassembling the channel into 2 channels, wherein one channel stores high-bit data, and the other channel stores low-bit data;
and (3.3) calculating the image uniformity of the channel image data obtained after the processing in the step (3.2), and dividing the channel image data into a more uniform image channel and an image channel with poor uniformity after determining the uniformity index.
Further, an image channel has M rows and N columns, and the average value of the pixels in the rows is uiThe process for determining the uniformity evaluation index is as follows:
autocorrelation coefficients rho of adjacent rows of pixels within an image channelxThe concrete formula is as follows:
mean value of image channel row autocorrelation coefficients
Autocorrelation coefficients rho of adjacent columns of pixels within an image channelyIn particularThe formula is as follows:
mean value of autocorrelation coefficients of image channel row
Combining image channel row autocorrelation coefficient mean values
And image channel column autocorrelation coefficient mean
Defining an image uniformity evaluation index psr, wherein a specific formula is as follows:
further, the step (4) specifically includes:
and encoding the more uniform image channel by adopting a FILF compressor, calculating gray value distribution of the image channel with poor uniformity, fitting the gray value distribution by using a Gaussian function, storing the fitted Gaussian function parameter, and entropy encoding the channel image by combining an arithmetic encoder and the fitted Gaussian function parameter.
Further, the step (5) specifically comprises:
decoding the compressed hyperspectral image channel by channel, reading a fitted Gaussian function parameter for an image channel to be decoded, which is coded by an arithmetic coder, and decoding image data of the channel by combining the arithmetic coder and the fitted Gaussian function parameter; the image channel to be decoded, which is encoded using the FILF compressor, is decoded by the FILF compressor.
Further, the step (6) specifically includes:
reading the transformation matrix obtained in the step (2), and performing reversible integer inverse transformation on the image without the inter-spectral correlation;
checking the record of the disassembled channel, and combining the disassembled channel according to the high order and the low order;
and reading the minimum value of the image data of each channel after combination, adding the minimum value to the corresponding channel, and recovering the original hyperspectral image before compression.
According to the traditional method for compressing the hyperspectral images by directly using the visible light compressor, three channels adjacent to the hyperspectral images are usually grouped and compressed, and although the visible light compressor also has the capacity of eliminating the inter-spectrum redundancy in the groups, the inter-spectrum redundancy information between the groups cannot be eliminated. There are many methods for removing the inter-spectrum correlation, but most of the transformations are real number transformations, which are not suitable for lossless compression of images, and can only be used for lossy compression of images. In addition to requiring the algorithm to be an integer transform, the algorithm must be reversible in order to achieve lossless compression of the data. At present, the reversible integer transformation method only comprises reversible integer KLT transformation and integer wavelet transformation, but the integer wavelet transformation belongs to frequency domain transformation and is more suitable for eliminating the spatial information correlation in an image spectrum, the inter-spectrum correlation can not be completely removed, while the integer KLT transformation belongs to orthogonal transformation, the inter-spectrum data are orthogonal after transformation, and the capability of eliminating the inter-spectrum correlation is stronger; the FILF compressor is selected because the compressor is the compressor with the best compression capability in the visible light image field at present, and the performance of the FILF compressor exceeds that of popular compressors such as PNG, JPEG-LS and JPEG 2000. No matter what spectral band of the hyperspectral image, the imaging wave bands of the adjacent channel images are very close, and generally, stronger inter-spectral correlation exists.
According to another aspect of the invention, a lossless compression system suitable for full-spectrum hyperspectral images is provided, which is characterized by comprising a preprocessing module, an inter-spectrum correlation removing module, an image uniformity evaluation module, a FILF encoder and arithmetic encoder, a FILF decoder and arithmetic decoder, and an image restoration module;
the preprocessing module is used for preprocessing the original hyperspectral image and correcting nonlinear factors caused by imaging;
the inter-spectrum correlation removal module is used for performing reversible integer transformation on the preprocessed hyperspectral images, removing inter-spectrum correlation and storing a transformation matrix;
the image uniformity evaluation module is used for calculating the image uniformity of the hyperspectral image subjected to reversible integer transformation channel by channel, and dividing the hyperspectral image subjected to reversible integer transformation into a more uniform image channel and an image channel with poor uniformity after determining uniformity evaluation indexes;
the FILF encoder and the arithmetic encoder are respectively used for encoding the more uniform image channel and the image channel with poor uniformity, so that lossless compression of the hyperspectral image is realized;
the FILF decoder and the arithmetic decoder are used for decoding the compressed hyperspectral images channel by channel, and the images without the inter-spectral correlation are decoded by adopting arithmetic encoder coding and adopting the image channel to be decoded coded by the FILF compressor in a corresponding decoding mode;
and the image recovery module is used for reading the transformation matrix obtained by the inter-spectrum correlation removal module, performing reversible integer inverse transformation on the image without the inter-spectrum correlation, and recovering the original hyperspectral image before compression.
Further, the image uniformity evaluation module includes:
the high and low bit splitting unit is used for calculating and solving the minimum value of the hyperspectral image subjected to reversible integer transformation channel by channel and storing the hyperspectral image, then subtracting the minimum value of the corresponding channel by channel, so that the channel image data is a positive integer which is more than or equal to 0, checking whether the maximum value overflows channel by channel, and if the channel image data overflows, splitting the channel into 2 channels, wherein one channel stores high bit data and the other channel stores low bit data;
and the uniformity determining unit is used for calculating the image uniformity of the obtained channel image data, and dividing the obtained channel image data into a more uniform image channel and an image channel with poor uniformity after determining the uniformity index.
Further, a FILF compressor is adopted for encoding a more uniform image channel, gray value distribution is calculated for an image channel with poor uniformity, a Gaussian function is used for fitting the gray value distribution, fitted Gaussian function parameters are stored, and entropy encoding is carried out on the channel image by combining an arithmetic encoder and the fitted Gaussian function parameters;
decoding the compressed hyperspectral image channel by channel, reading a fitted Gaussian function parameter for an image channel to be decoded, which is coded by an arithmetic coder, and decoding image data of the channel by combining the arithmetic coder and the fitted Gaussian function parameter; the image channel to be decoded, which is encoded using the FILF compressor, is decoded by the FILF compressor.
Through the technical scheme, compared with the prior art, the invention has the following beneficial effects:
1. the spectrum-space dimension combined hyperspectral image lossless compression method is suitable for full-waveband hyperspectral images, removes strong correlation among all the spectral bands of the hyperspectral images through reversible integer KLT transformation, reduces the information entropy of the hyperspectral images, eliminates the information redundancy among the spectrums, and improves the performance of a FILF compressor; and the problem of possible integer overflow is solved, and the robustness of the FILF compressor is improved.
2. The FILF compressor adopted by the invention belongs to a complex compression system, has strong compression performance, but has high complexity of an internal algorithm and long time consumption; the arithmetic coder belongs to a simple compression system, and has high speed but weak compression capability. In order to shorten the compression time on the premise of not losing the compression performance as much as possible, the invention adopts a strategy of flexibly switching between a FILF compressor and an arithmetic coder according to the image uniformity degree, and the FILF compressor is selected for uniform images to improve the compression capability; but the FILF compressor performs comparably to an arithmetic encoder for noisy non-uniform images, where a faster arithmetic encoder is chosen to shorten the compression time.
Drawings
FIG. 1 is a flow chart of the present invention providing improved FILF practical lossless compression of hyperspectral images;
FIG. 2 is a disclosed hyperspectral data Salinas image provided by the invention;
fig. 3(a) is a diagram of channel 1 of the KLT transformed image Salinas provided by the present invention;
fig. 3(b) is a 2 nd channel diagram of the KLT transformed image Salinas provided by the present invention;
fig. 3(c) is a diagram of channel 203 of the KLT transformed image Salinas provided by the present invention;
fig. 3(d) is a diagram of channel 204 of the KLT transformed image Salinas provided by the present invention;
FIG. 4(a) is a graph of correlation between adjacent spectral bands prior to transformation provided by the present invention;
FIG. 4(b) is a graph of correlation between adjacent spectral bands after transformation according to the present invention;
FIG. 5 is a trend graph of the maximum dynamic range of pixel values of each channel after Salinas processing according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The invention provides a method for removing the correlation between hyperspectral image spectrums, which is different from common real number domain KLT analysis. Therefore, the hyperspectral image lossless compression method is particularly suitable for application scenes of hyperspectral image data lossless compression.
As shown in FIG. 1, the invention provides a spectrum-space dimension combined hyperspectral image lossless compression method, which comprises the following steps:
(1) preprocessing an original hyperspectral image, and correcting nonlinear factors such as image deformation and response drift caused by imaging;
(2) reversible integer transformation is carried out on the preprocessed hyperspectral images, the correlation among spectrums is removed, and a transformation matrix is stored;
(3) calculating the image uniformity of the hyperspectral images subjected to reversible integer transformation channel by channel, and dividing the hyperspectral images into more uniform image channels and image channels with poor uniformity after determining uniformity evaluation indexes;
(4) the more uniform image channels are coded by a FILF compressor, the image channels with poor uniformity are coded by an arithmetic coder, and lossless compression of the hyperspectral image is realized;
(5) decoding the compressed hyperspectral images channel by channel, and decoding the images without the inter-spectral correlation by adopting an arithmetic encoder to encode and adopting an image channel to be decoded encoded by a FILF compressor in a corresponding decoding mode;
(6) and (3) reading the transformation matrix obtained in the step (2), and performing reversible integer inverse transformation on the image without the inter-spectrum correlation to restore the original hyperspectral image before compression.
The invention also provides a lossless compression system suitable for the full-spectrum hyperspectral image, which is characterized by comprising a preprocessing module, an inter-spectrum correlation removing module, an image uniformity evaluation module, a FILF encoder, an arithmetic encoder, a FILF decoder, an arithmetic decoder and an image recovery module;
the preprocessing module is used for preprocessing an original hyperspectral image and correcting nonlinear factors such as image deformation and response drift caused by imaging;
the inter-spectrum correlation removal module is used for performing reversible integer transformation on the preprocessed hyperspectral images, removing inter-spectrum correlation and storing a transformation matrix;
the image uniformity evaluation module is used for calculating the image uniformity of the hyperspectral image subjected to reversible integer transformation channel by channel, and dividing the hyperspectral image subjected to reversible integer transformation into a more uniform image channel and an image channel with poor uniformity after determining uniformity evaluation indexes;
the FILF encoder and the arithmetic encoder are respectively used for encoding the more uniform image channel and the image channel with poor uniformity, so that lossless compression of the hyperspectral image is realized;
the FILF decoder and the arithmetic decoder are used for decoding the compressed hyperspectral images channel by channel, and the images without the inter-spectral correlation are decoded by adopting arithmetic encoder coding and adopting the image channel to be decoded coded by the FILF compressor in a corresponding decoding mode;
and the image recovery module is used for reading the transformation matrix obtained by the inter-spectrum correlation removal module, performing reversible integer inverse transformation on the image without the inter-spectrum correlation, and recovering the original hyperspectral image before compression.
Specifically, the image uniformity evaluation module includes:
the high and low bit splitting unit is used for calculating and solving the minimum value of the hyperspectral image subjected to reversible integer transformation channel by channel and storing the hyperspectral image, then subtracting the minimum value of the corresponding channel by channel, so that the channel image data is a positive integer which is more than or equal to 0, checking whether the maximum value overflows channel by channel, and if the channel image data overflows, splitting the channel into 2 channels, wherein one channel stores high bit data and the other channel stores low bit data;
and the uniformity determining unit is used for calculating the image uniformity of the obtained channel image data, and dividing the obtained channel image data into a more uniform image channel and an image channel with poor uniformity after determining the uniformity index.
Specifically, a FILF compressor is adopted for encoding a more uniform image channel, gray value distribution is calculated for an image channel with poor uniformity, a Gaussian function is used for fitting the gray value distribution, fitted Gaussian function parameters are stored, and entropy encoding is carried out on the channel image by combining an arithmetic encoder and the fitted Gaussian function parameters;
decoding the compressed hyperspectral image channel by channel, reading a fitted Gaussian function parameter for an image channel to be decoded, which is coded by an arithmetic coder, and decoding image data of the channel by combining the arithmetic coder and the fitted Gaussian function parameter; the image channel to be decoded, which is encoded using the FILF compressor, is decoded by the FILF compressor.
The invention improves the system compression performance and shortens the compression time by removing the correlation among spectrums by using integer KLT and flexibly switching the strategy of a compressor, and in a specific embodiment, the invention comprises the following main steps:
(1) according to the corresponding preprocessing flow of the imaging equipment, preprocessing is carried out on the original hyperspectral image data, and nonlinear factors such as image deformation and response drift caused by the imaging instrument are corrected. Specifically, the input image is a public hyperspectral data image Salinas, as shown in fig. 2, the image Salinas is a 16-bit grayscale map with 204 channels and a size of 512 × 217, and the following experimental results are all completed under the condition. The original image, Salinas, was preprocessed using the ENVI software tool and saved as Salinas _ corrected. The preprocessing step is very critical, and the correlation between data spectrums which are not preprocessed is poor, so that the compression performance of the algorithm is reduced.
(2) And performing reversible integer KLT forward transformation on the preprocessed hyperspectral image Salinas _ corrected, removing the correlation among spectrums, and storing a transformation matrix.
Specifically, first, a Pearson correlation matrix Coeff between Salinas _ corrected spectra of the image is calculated:
where ρ (p)i，pj) The Pearson correlation coefficients of the ith and jth spectral bands are expressed, and the specific calculation formula is as follows:
wherein, Cov (p)i，pj) Denotes the covariance between spectral bands, D (p)i)、D(pj) Is the variance of the ith and jth spectral bands.
And then, carrying out principal component analysis on the inter-spectrum Pearson correlation matrix Coeff to obtain a transformation matrix A.
The transformation matrix A can only realize the transformation from real number to real number, and the transformation matrix A is obtained by decomposition*PLUS implements integer to integer mapping, where L and U are unit lower and upper triangular arrays, respectively, and P is permutationMatrix, S is a reversible matrix of unit elements. The specific decomposition process of A is as follows:
let A be the nonsingular matrix that satisfies the decomposition condition:
first, there is a row transformation matrix P1Make it
First element of the Nth column
Is not zero, s can be found1Make it
Namely, it is
Thus, a unit element invertible matrix S is constructed1：
Then there is
Then, a traditional Gaussian elimination method is adopted to define an elementary Gaussian matrix:
then there are:
repeating the above process N-1 times to obtain:
LN-1PN-1…L2P2L1P1AS1S2…SN-1＝DRU
wherein k is 1, 2, … N-1; dRIs a twiddle factor, DR＝±1
The transformation matrix a can be finally decomposed into the following equation:
A＝PLUS
wherein:
P＝(PN-1…P2P1)T
since L is a lower triangular matrix, U is an upper triangular matrix, and only the last row of the S matrix has valid data, the LUSs can be combined into a real matrix to be stored.
And (3) sequentially multiplying the high-spectrum image data by the S, U, L and P transformation matrixes, and rounding after multiplying the matrixes each time to realize integer KLT transformation, wherein a specific formula is as follows:
Y＝P*round(L*round(U*round(S*data)))
where round denotes the rounding operation.
(3) And solving and storing the minimum value of the high-spectrum data subjected to KLT transformation channel by channel, and subtracting the minimum value of the corresponding channel by channel to enable the image data to be a positive integer larger than or equal to 0.
For the specific image Salinas _ corrected, channels 1, 2, 203 and 204 of the transformed image are extracted as shown in fig. 3(a), fig. 3(b), fig. 3(c) and fig. 3(d), respectively; to illustrate the effect of the reversible integer KLT transform to remove the inter-spectral correlation, the correlations of adjacent spectral bands before and after the transform are calculated separately, as shown in fig. 4(a) and 4(b), and as can be seen from fig. 4, the inter-spectral correlation after the KLT transform is substantially eliminated.
(4) Checking whether the maximum overflow problem exists in the data obtained after the processing in the step (3) channel by channel, if the channel image data has overflow, disassembling the channel data into 2 channels, wherein one channel stores high-order data, the other channel stores low-order data, the serial number of the disassembled data channel is recorded, and the data is checked repeatedly until the maximum overflow problem does not exist.
For a specific image, Salinas _ corrected, the maximum value after KLT transform can be calculated by the following formula:
where BIT represents the number of image BITs and n represents the number of image spectral segments. Knowing that the image Salinas _ corrected is 16 bits, with 204 spectral bins, a maximum value of 1872054 can be calculated. Therefore, splitting up to 2 16-bit lanes can solve the maximum overflow problem.
The maximum dynamic range of the processed pixel values of each channel is shown in fig. 5 below.
(5) And (4) calculating the image uniformity index psr of the data obtained after the processing in the step (4) channel by channel.
Calculating autocorrelation coefficients rho of pixels in adjacent rows in an imagexThe concrete formula is as follows:
calculating the mean value of the image row autocorrelation coefficients
Calculating autocorrelation coefficients rho of adjacent column pixels in an imageyThe concrete formula is as follows:
calculating the mean value of the autocorrelation coefficients of the image rows
Combining image line autocorrelation coefficient means
And image column autocorrelation coefficient mean
Defining an image uniformity evaluation index psr, wherein a specific formula is as follows:
(6) encoding the image channel with strong uniformity by adopting a FILF compressor; and calculating gray value distribution of image channel data with poor uniformity, fitting the gray value distribution by using a Gaussian function, storing fitted Gaussian function parameters, and performing entropy coding on the channel image data by combining an arithmetic coder and the fitted Gaussian distribution function.
For a specific image, Salinas _ corrected, it is generally considered that a non-uniform image with a uniformity index psr less than 0.3.
The hyperspectral image decoding comprises the following steps:
(7) and decoding the hyperspectral image data channel by channel. If the channel to be decoded is compressed by an arithmetic coder, the parameters of the fitted Gaussian function are read firstly, and then the image data of the channel is decoded by combining the arithmetic coder and the fitted Gaussian distribution function.
(8) And checking the record of the disassembled channel, and combining the disassembled channel according to the high order and low order.
(9) The minimum value of the image data of each channel is read and added to the corresponding channel.
(10) And (4) reading the reversible integer KLT transformation matrix, and carrying out KLT inverse transformation on the data obtained after the processing in the step (9). The specific transformation flow is as follows:
the transformation matrix PLUS is first read.
Assuming that the high spectral data after KLT transformation is Y and the data before the transformation is X, in order to recover X, the following equations are solved in sequence:
PLUSX＝Y
since P is an elementary matrix, which is itself an integer transform, luxx can be solved:
LUSX＝PTY
l is the lower triangular matrix, the inverse transformed integer is implemented as a top-down calculation:
at this time, USX can be solved.
Similarly, U is an upper triangular matrix, and the inverse transformed integer is calculated from bottom to top, so that SX can be solved.
Similarly, S is a lower triangular matrix, and the hyperspectral image data X before transformation can be solved by the solving method.
To this end, the entire compression and decompression flow is as described above.
For comparing the effects of the present invention, one of the Salinas _ corrected images is directly compressed by a FILF compressor, and the other image is directly compressed by the method of the present invention, where the compression ratio CR is defined as the ratio of the size after compression to the size before compression, and the final test results are shown in table 1 below.
TABLE 1
| Compression Ratio (CR) | Compression time (seconds) | |
| FILF only | 3.07 | 84 |
| The method of the invention | 3.39 | 52 |
From the above table, the image compression ratio can be effectively improved, and is improved by 0.3 compared with the FILF method; at the same time, the compression time is reduced by nearly half.
The method firstly preprocesses an input image, and then removes the correlation between spectrums by using reversible integer KLT forward transform. Then, the uniformity index psr is calculated on an image channel-by-image channel basis, and the compressor is flexibly switched according to the image uniformity. By executing the method, the inter-spectrum correlation of the hyperspectral image can be effectively removed, so that the FILF compression performance is greatly improved. In addition, the invention can be flexibly switched to an arithmetic coder, thereby further reducing the algorithm complexity and improving the algorithm compression speed.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (6)
1. A spectrum-space dimension combined hyperspectral image lossless compression method is characterized by comprising the following steps:
(1) preprocessing an original hyperspectral image, and correcting nonlinear factors caused by imaging;
(2) reversible integer transformation is carried out on the preprocessed hyperspectral images, the correlation among spectrums is removed, and a transformation matrix is stored; the method specifically comprises the following steps:
(2.1) calculating a Pearson correlation matrix Coeff among the preprocessed hyperspectral image data spectrums:
where ρ (p)i，pj) The Pearson correlation coefficients of the ith and jth spectral bands are expressed, and the specific calculation formula is as follows:
wherein, Cov (p)i，pj) Denotes the covariance between spectral bands, D (p)i)、D(pj) Is the variance of the ith and jth spectral bands;
(2.2) carrying out principal component analysis on the Pearson correlation matrix Coeff to obtain a transformation matrix A, and decomposing the transformation matrix A into the following formula:
A＝PLUS
wherein:
P＝(PN-1…P2P1)T
U＝±LN-1PN-1…L2P2L1P1AS1S2…SN-1
the transformation matrix A can only realize the transformation from real numbers to real numbers, and the decomposed transformation matrix realizes the mapping from integers to integers;
(3) calculating image uniformity of the hyperspectral images subjected to reversible integer transformation channel by channel, and dividing the hyperspectral images after determining uniformity evaluation indexes into a first image channel and a second image channel which are different in uniformity; the method specifically comprises the following steps:
(3.1) calculating and saving the minimum value of the hyperspectral images subjected to reversible integer transformation channel by channel, and subtracting the minimum value of the corresponding channel by channel so that the channel image data are positive integers which are more than or equal to 0;
(3.2) checking channel-by-channel whether the channel image data obtained after the processing of the step (3.1) has the maximum overflow problem, and if the channel image data has overflow, disassembling the channel into 2 channels, wherein one channel stores high-bit data, and the other channel stores low-bit data;
(3.3) calculating the image uniformity of the channel image data obtained after the processing in the step (3.2), and dividing the channel image data into a first image channel and a second image channel with different uniformity after determining the uniformity index;
wherein one image channel has M rows and N columns, and the average value of the pixels in the rows is uiThe process for determining the uniformity evaluation index comprises the following steps:
autocorrelation coefficients rho of adjacent rows of pixels within an image channelxThe concrete formula is as follows:
mean value of image channel row autocorrelation coefficients
Autocorrelation coefficients rho of adjacent columns of pixels within an image channelyThe concrete formula is as follows:
mean value of autocorrelation coefficients of image channel row
Combining image channel row autocorrelation coefficient mean values
And image channel column autocorrelation coefficient mean
Defining an image uniformity evaluation index psr, wherein a specific formula is as follows:
(4) the first image channel is coded by adopting a FILF compressor, and the second image channel is coded by adopting an arithmetic coder, so that lossless compression of the hyperspectral image is realized;
(5) decoding the compressed hyperspectral images channel by channel, and decoding the images without the inter-spectral correlation by adopting an arithmetic encoder to encode and adopting an image channel to be decoded encoded by a FILF compressor in a corresponding decoding mode;
(6) and (3) reading the transformation matrix obtained in the step (2), and performing reversible integer inverse transformation on the image without the inter-spectral correlation to restore the original hyperspectral image before compression.
2. The compression method according to claim 1, characterized in that said step (4) comprises in particular:
and encoding the first image channel by adopting a FILF compressor, calculating gray value distribution of the second image channel, fitting the gray value distribution by using a Gaussian function, storing the fitted Gaussian function parameters, and entropy encoding the channel image by combining the arithmetic encoder and the fitted Gaussian function parameters.
3. The compression method according to claim 1, characterized in that said step (5) comprises in particular:
decoding the compressed hyperspectral image channel by channel, reading a fitted Gaussian function parameter for an image channel to be decoded, which is coded by an arithmetic coder, and decoding channel image data by combining the arithmetic coder and the fitted Gaussian function parameter; the image channel to be decoded, which is encoded using the FILF compressor, is decoded by the FILF compressor.
4. The compression method according to claim 1, characterized in that said step (6) comprises in particular:
reading the transformation matrix obtained in the step (2), and performing reversible integer inverse transformation on the image without the inter-spectral correlation;
checking the record of the disassembled channel, and combining the disassembled channel according to the high order and the low order;
and reading the minimum value of the image data of each channel after combination, adding the minimum value to the corresponding channel, and recovering the original hyperspectral image before compression.
5. The spectrum-space dimension combined hyperspectral image lossless compression system is characterized by comprising a preprocessing module, a inter-spectrum correlation removing module, an image uniformity evaluation module, a FILF (first-class image frequency) encoder and arithmetic encoder, a FILF decoder and arithmetic decoder and an image restoration module;
the preprocessing module is used for preprocessing an original hyperspectral image and correcting nonlinear factors caused by imaging;
the inter-spectrum correlation removal module is used for performing reversible integer transformation on the preprocessed hyperspectral images, removing inter-spectrum correlation and storing a transformation matrix; the method specifically comprises the following steps: calculating a Pearson correlation matrix Coeff among the preprocessed hyperspectral image data spectrums:
where ρ (p)i，pj) The Pearson correlation coefficients of the ith and jth spectral bands are expressed, and the specific calculation formula is as follows:
wherein, Cov (p)i，pj) Denotes the covariance between spectral bands, D (p)i)、D(pj) Is the variance of the ith and jth spectral bands;
performing principal component analysis on the Pearson correlation matrix Coeff to obtain a transformation matrix A, and decomposing the transformation matrix A into the following formula:
A＝PLUS
wherein:
P＝(PN-1…P2P1)T
U＝±LN-1PN-1…L2P2L1P1AS1S2…SN-1
the transformation matrix A can only realize the transformation from real numbers to real numbers, and the decomposed transformation matrix realizes the mapping from integers to integers;
the image uniformity evaluation module is used for calculating the image uniformity of the hyperspectral image subjected to reversible integer transform channel by channel, and dividing the hyperspectral image subjected to reversible integer transform into a first image channel and a second image channel with different uniformity after determining uniformity evaluation indexes; the image uniformity evaluation module comprises:
the high and low bit splitting unit is used for calculating and solving the minimum value of the hyperspectral image subjected to reversible integer transformation channel by channel and storing the hyperspectral image, then subtracting the minimum value of the corresponding channel by channel, so that the channel image data is a positive integer which is more than or equal to 0, checking whether the maximum value overflows channel by channel, and if the channel image data overflows, splitting the channel into 2 channels, wherein one channel stores high bit data and the other channel stores low bit data;
the uniformity determining unit is used for calculating the image uniformity of the obtained channel image data, and dividing the channel image data into a first image channel and a second image channel with different uniformity after determining the uniformity index; wherein one image channel has M rows and N columns, and the average value of the pixels in the rows is uiThe process for determining the uniformity evaluation index comprises the following steps:
autocorrelation coefficients rho of adjacent rows of pixels within an image channelxThe concrete formula is as follows:
mean value of image channel row autocorrelation coefficients
Autocorrelation coefficients rho of adjacent columns of pixels within an image channelyThe concrete formula is as follows:
mean value of autocorrelation coefficients of image channel row
Combining image channel row autocorrelation coefficient mean values
And image channel column autocorrelation coefficient mean
Defining an image uniformity evaluation index psr, wherein a specific formula is as follows:
the FILF encoder and the arithmetic encoder are respectively used for encoding the first image channel and the second image channel to realize lossless compression of the hyperspectral image;
the FILF decoder and the arithmetic decoder are used for decoding the compressed hyperspectral images channel by channel, and the images without the inter-spectral correlation are decoded by adopting arithmetic encoder coding and adopting the image channel to be decoded coded by the FILF compressor in a corresponding decoding mode;
the image recovery module is used for reading the transformation matrix obtained by the inter-spectrum correlation removal module, and performing reversible integer inverse transformation on the image without the inter-spectrum correlation to recover the original hyperspectral image before compression.
6. The system of claim 5, wherein the first image channel is encoded using a FILF compressor, the second image channel is encoded using a gray value distribution, the gray value distribution is fitted using a Gaussian function, the fitted Gaussian function parameters are stored, and the channel image is entropy encoded using the arithmetic encoder and the fitted Gaussian function parameters;
decoding the compressed hyperspectral image channel by channel, reading a fitted Gaussian function parameter for an image channel to be decoded, which is coded by an arithmetic coder, and decoding channel image data by combining the arithmetic coder and the fitted Gaussian function parameter; the image channel to be decoded, which is encoded using the FILF compressor, is decoded by the FILF compressor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911420225.1A CN111107360B (en) | 2019-12-31 | 2019-12-31 | Spectrum-space dimension combined hyperspectral image lossless compression method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911420225.1A CN111107360B (en) | 2019-12-31 | 2019-12-31 | Spectrum-space dimension combined hyperspectral image lossless compression method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111107360A CN111107360A (en) | 2020-05-05 |
| CN111107360B true CN111107360B (en) | 2020-12-08 |
Family
ID=70425866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911420225.1A Active CN111107360B (en) | 2019-12-31 | 2019-12-31 | Spectrum-space dimension combined hyperspectral image lossless compression method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111107360B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113961886B (en) * | 2021-09-28 | 2024-04-19 | 西安交通大学 | Reversible information hiding method and system for hyperspectral image |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101754008A (en) * | 2008-12-10 | 2010-06-23 | 解成俊 | Uniform parametric three-band spectral integer reversible transformation |
| CN101883274B (en) * | 2009-05-08 | 2013-05-22 | 中国科学院沈阳自动化研究所 | Spatial-spectral associated prediction-based hyperspectral image lossless compression method |
| US8655091B2 (en) * | 2012-02-24 | 2014-02-18 | Raytheon Company | Basis vector spectral image compression |
| CN104270642B (en) * | 2014-10-22 | 2017-07-28 | 哈尔滨工业大学 | A kind of lossless compression method for high spectrum image chosen based on RKLT and principal component |
| CN107730447A (en) * | 2017-10-31 | 2018-02-23 | 北京信息科技大学 | A kind of Hyperspectral imagery processing method, apparatus and system |
-
2019
- 2019-12-31 CN CN201911420225.1A patent/CN111107360B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN111107360A (en) | 2020-05-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109889839B (en) | Region-of-interest image coding and decoding system and method based on deep learning | |
| CN111641832B (en) | Encoding method, decoding method, device, electronic device and storage medium | |
| Kozhemiakin et al. | Image quality prediction for DCT-based compression | |
| Kumar et al. | A review: DWT-DCT technique and arithmetic-Huffman coding based image compression | |
| CN107124612B (en) | Method for compressing high spectrum image based on distributed compression perception | |
| CN113962882A (en) | JPEG image compression artifact eliminating method based on controllable pyramid wavelet network | |
| CN115361559A (en) | Image encoding method, image decoding method, image encoding device, image decoding device, and storage medium | |
| CN111107360B (en) | Spectrum-space dimension combined hyperspectral image lossless compression method and system | |
| CN112150356A (en) | Single compressed image super-resolution reconstruction method based on cascade framework | |
| CN108182712B (en) | Image processing method, device and system | |
| Sadkhan | A proposed image compression technique based on DWT and predictive techniques | |
| Alsayyh et al. | A Novel Fused Image Compression Technique Using DFT, DWT, and DCT. | |
| CN106331719A (en) | K-L transformation error space dividing based image data compression method | |
| Anasuodei et al. | An enhanced satellite image compression using hybrid (DWT, DCT and SVD) algorithm | |
| Awad et al. | Improving Reconstructed Image Quality via Hybrid Compression Techniques. | |
| CN110930466B (en) | Hyperspectral self-adaptive compressive sensing method for BOIs with arbitrary shape | |
| Panda et al. | Competency assessment of image compression in the lossy and lossless domain | |
| Kunwar | Strategies in JPEG compression using Convolutional Neural Network (CNN) | |
| Rani et al. | Medical image compression using DCT with entropy encoding and huffman on MRI brain images | |
| Narasimhulu et al. | Hybrid LWT and DCT based High-Quality Image Compression | |
| US20050089236A1 (en) | Image transformation method and apparatus, and storage medium | |
| CN114882133B (en) | Image coding and decoding method, system, device and medium | |
| CN114998457B (en) | Image compression method, image decompression method, related device and readable storage medium | |
| Wiewel | FPGA implementation of an energy-efficient real-time image compression algorithm for the EIVE satellite mission | |
| Li et al. | BPG-Based Lossy Compression of Three-Channel Remote Sensing Images with Visual Quality Control |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |