Low-complexity lossy compression of hyperspectral images via informed quantization

Convolution Neural Network based lossy compression of hyperspectral images

Hyperspectral image compression has become increasingly important in data exploitation because of enormous data volumes and high redundancy provided by hundreds of contiguous spectral channels. Since a hyperspectral image can be viewed as a 3-dimensional (3D) image cube, many efforts have been devoted to extending 2D image compression techniques to perform 3D image compression on hyperspectral image cubes. Unfortunately, some major issues generally encountered in hyperspectral data exploitation at low or very low-bit rate compression, for example, subpixels and mixed pixels which do not occur in traditional pure pixel-based image compression are often overlooked in such a 2D-to-3D compression. Accordingly, a direct application of 2D-to-3D compression techniques to hyperspectral image cubes without taking precaution may result in significant loss of crucial spectral information provided by subtle substances such as small objects, anomalies during low bit-rate lossy compression. This paper takes a rather different view by investigating lossy hyperspectral compression from a perspective of exploring spectral information, referred to as exploitation-based lossy compression and further develops spectral/spatial hyperspectral image compression to effectively preserve crucial and vital spectral information of objects which are generally missed by commonly used mean-squared error (MSE) or signal-to-noise ratio (SNR)-based compression techniques when lossy compression is performed at low bit rates. In order to demonstrate advantages of the proposed spectral/spatial compression approach applications of subpixel target detection and mixed pixel analysis are used for experiments for performance evaluation.

Nonnegative tensor Tucker decomposition (NTD) in a transform domain (e.g., 2D-DWT, etc) has been used in the compression of hyper-spectral images because it can remove redundancies between spectrum bands and also exploit spatial correlations of each band. However, the use of a NTD has a very high computational cost. In this paper, we propose a low complexity NTD-based compression method of hyper-spectral images. This method is based on a pair-wise multilevel grouping approach for the NTD to overcome its high computational cost. The proposed method has a low complexity under a slight decrease of the coding performance compared to conventional NTD. We experimentally confirm this method, which indicates that this method has the less processing time and keeps a better coding performance than the case that the NTD is not used. The proposed approach has a potential application in the loss compression of hyper-spectral or multi-spectral images

Hyperspectral (HS) image is a three dimensional data image where the 3rd dimension carries the wealth of spectrum information. HS image compression is one of the areas that has attracted increasing attention for big data processing and analysis. HS data has its own distinguishing feature which differs with video because without motion, also different with a still image because of redundancy along the wavelength axis. The prediction based method is playing an important role in the compression and research area. Reflectance distribution of HS based on our analysis indicates that there is some nonlinear relationship in intra-band. The Multilayer Propagation Neural Networks (MLPNN) with backpropagation training are particularly well suited for addressing the approximation function. In this paper, an MLPNN based predictive image compression method is presented. We propose a hybrid Adaptive Prediction Mechanism (APM) with MLPNN model (APM-MLPNN). MLPNN is trained to predict the succeeding bands by using current band information. The purpose is to explore whether MLPNN can provide better image compression results in HS images. Besides, it uses less computation cost than a deep learning model so we can easily validate the model. We encoded the weights vector and the bias vector of MLPNN as well as the residuals. That is the only few bytes it then sends to the decoder side. The decoder will reconstruct a band by using the same structure of the network. We call it an MLPNN decoder. The MLPNN decoder does not need to be trained as the weights and biases have already been transmitted. We can easily reconstruct the succeeding bands by the MLPNN decoder. APM constrained the correction offset between the succeeding band and the current spectral band in order to prevent HS image being affected by large predictive biases. The performance of the proposed algorithm is verified by several HS images from Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) reflectance dataset. MLPNN simulation results can improve prediction accuracy; reduce residual of intra-band with high compression ratio and relatively lower bitrates.

To review the literature on lossy compression in dental radiography and to discuss the importance and suitability of the methodology used for evaluation of image compression. A search of Medline (from 1966 to October 2004) was undertaken with the search expression "(Radiography, dental) and compression". Inclusion criterion was that the reference should be evaluating the effect of lossy image compression on diagnostic accuracy. For all included studies, information in relation to mode of image acquisition, image content, image compression, image display, and method of image evaluation was extracted. 12 out of 32 papers were included in the review. The design of these 12 studies was found to vary considerably. Parameters used to express the degree of information loss (DIL) were either or both compression ratio (CR) and compression level (CL). The highest acceptable CR reported in the studies ranged from 3.6% to 15.4%. Furthermore, different CR values were proposed even for the same diagnostic task, for example, for caries diagnosis CR ranged from 6.2% to 11.1%. Lossy image compression can be used in clinical radiology if it does not conflict with national law. However, the acceptable DIL is difficult to express and standardize. CR is probably not suitable to express DIL, because it is image content dependent. CL is also probably not suitable to express DIL because of the lack of compression software standardization.

There has recently been a strong interest towards low-complexity approaches for hyperspectral image compression, also driven by the standardization activities in this area and by the new hyperspectral missions that have been deployed. This chapter overviews the state-of-the-art of lossless and near-lossless compression of hyperspectral images, with a particular focus on approaches that comply with the requirements typical of real-world mission, in terms of low complexity and memory usage, error resilience and hardware friendliness. In particular, a very simple lossless compression algorithm is described, which is based on block-by-block prediction and adaptive Golomb coding, can exploit optimal band ordering, and can be extended to near-lossless compression. We also describe the results obtained with a hardware implementation of the algorithm. The compression performance of this algorithm is close to the state-of-the-art, and its low degree of complexity and memory usage, along with the possibility to compress data in parallel, make it a very good candidate for onboard hyperspectral image compression.

The compression of hyperspectral images (HSIs) has recently become a very attractive issue for remote sensing applications because of their volumetric data. In this paper, an efficient method for hyperspectral image compression is presented. The proposed algorithm, based on Discrete Wavelet Transform and Tucker Decomposition (DWT-TD), exploits both the spectral and the spatial information in the images. The core idea behind our proposed technique is to apply TD on the DWT coefficients of spectral bands of HSIs. We use DWT to effectively separate HSIs into different sub-images and TD to efficiently compact the energy of sub-images. We evaluate the effect of the proposed method on real HSIs and also compare the results with the well-known compression methods. The obtained results show a better performance of the proposed method. Moreover, we show the impact of compression HSIs on the supervised classification and linear unmixing.

The compression of hyperspectral images becomes recently very attractive issue for remote sensing applications because of the volumetric data. In this paper, an efficient method for hyperspectral image compression is presented based on Tucker Decomposition (TD) and Discrete Wavelet Transform (DWT). The core idea behind our proposed technique is to apply TD on the DWT coefficients of spectral bands of hyperspectral images. Our method not only exploits redundancies between bands but also uses spatial correlation of every image band. Simulation results applied on the real hyperspectral images show a remarkable compression ratio and quality.

Depending on the mammoth size of the hyperspectral image, the wavelet transform-based compression algorithm has achieved impressive performance in the compression of hyperspectral images. Set partition wavelet transform hyperspectral image compression algorithms have superior performance than other transform compression algorithms, such as embeddedness, low coding complexity and high coding efficiency. Fractional wavelet-based zero memory set partitioned embedded block (ZM-SPECK) reduces the demand for transform memory and coding memory with at par transform complexity and high coding efficiency. But comparing every coefficient/block/set for each frequency frame with the current threshold is time-consuming. The present algorithm deals with the complexity and memory of the transform image coding algorithms. The block-based fractional wavelet filter delivers exact transform results like other wavelet transforms, but demands the least transform memory with at par wavelet transform complexity. With the employment of the low complexity zero memory set partitioned embedded block (LC-ZM-SPECK), the coding complexity of the compression algorithm is further reduced. The simulation results show that the proposed compression algorithm reduces the overall complexity by ~ 25% to other state-of-the-art compression algorithms and reduces the transform memory by ~ 40%, making it a suitable choice for the resource-constrained hyperspectral image sensors.

In order to improve the detection and recognition ability of multiscale hyperspectral images with weak edges, a lossless compression algorithm of multiscale hyperspectral images with weak edges based on improved deep learning is proposed. Using deep-level feature information extraction method to extract fine-grained features of cancer cells from multiscale hyperspectral images with weak edges, constructing 3D surface structure reconstruction model of multiscale hyperspectral images with weak edges, combining with variational Bayesian filter detection method to decompose multiscale color structure features of multiscale hyperspectral images with weak edges, and constructing gradually refined output structure model of multiscale hyperspectral images with weak edges, Fuzzy image coarse-scale neural network learning method is used to realize block matching and edge detection of multi-scale hyperspectral images with weak edges. Ultra-fine-grained features of multi-scale hyperspectral images with weak edges are extracted. Deep learning algorithm is used to adaptively optimize the lossless compression process of multi-scale hyperspectral images with weak edges, and the optimal design of lossless compression algorithm for multi-scale hyperspectral images with weak edges is realized. The simulation results show that the adaptive convergence of lossless compression of multi-scale weak edge hyperspectral images with this method is good, the feature resolution is strong, and the hyperspectral feature detection and recognition ability is good.

A common problem in applying lossy compression to a hyperspectral image is predicting its effect on spectral target detection performance. Recent work has shown that light amounts of lossy compression can remove noise in hyperspectral imagery that would otherwise bias a covariance-based spectral target detection algorithm’s background-normalized response to target samples. However, the detection performance of such an algorithm is a function of both the specific target of interest as well as the background, among other factors, and therefore sometimes lossy compression operating at a particular compression ratio (CR) will not negatively affect the detection of one target, while it will negatively affect the detection of another. To account for the variability in this behavior, we have developed a target-centric metric that guides the selection of a lossy compression algorithm’s CR without knowledge of whether or not the targets of interest are present in an image. Further, we show that this metric is correlated with the adaptive coherence estimator’s (ACE’s) signal to clutter ratio when targets are present in an image.

The high complexity of the reconstruction algorithm is the main bottleneck of the hyperspectral image (HSI) compression technology based on compressed sensing. Compressed sensing technology is an important tool for retrieving the maximum number of HSI scenes on the ground. However, the complexity of the compressed sensing algorithm is limited by the energy and hardware of spaceborne equipment. Aiming at the high complexity of compressed sensing reconstruction algorithm and low reconstruction accuracy, an equivalent model of the invertible transformation is theoretically derived by us in the paper, which can convert the complex invertible projection training model into the coupled dictionary training model. Besides, aiming at the invertible projection training model, the most competitive task-driven invertible projection matrix learning algorithm (TIPML) is proposed. In TIPML, we don’t need to directly train the complex invertible projection model, but indirectly train the invertible projection model through the training of the coupled dictionary. In order to improve the accuracy of reconstructed data, in the paper, the singular value transformation is proposed. It has been verified that the concentration of the dictionary is increased and that the expressive ability of the dictionary has not been reduced by the transformation. Besides, two-loop iterative training is established to improve the accuracy of data reconstruction. Experiments show that, compared with the traditional compressed sensing algorithm, the compressed sensing algorithm based on TIPML has higher reconstruction accuracy, and the reconstruction time is shortened by more than a hundred times. It is foreseeable that the TIPML algorithm will have a huge application prospect in the field of HSI compression.

Image compression, which is a type of data compression applied to digital images, has been a fundamental research topic for many decades. Recent image techniques produce very large amounts of data, which may make it prohibitive to storage and communications of image data without the use of compression. However, the traditional compression methods, such as JPEG, may introduce the compression artefact problems. Recently, deep learning has achieved great success in many computer vision tasks and is gradually being used in image compression. To solve the compression atrefact problem, in this paper, we present a lossy image compression architecture, which utilizes the advantages of the existing deep learning methods to achieve a high coding efficiency. We design a densely connected autoencoder structure for lossy image compression. Firstly, we design a densely autoencoder structure to get richer feature information from image which can be helpful for compression. Secondly, we design a U-net like network to decrease the distortion caused by compression. Finally, an improved binarizer is adopted to quantize the output of encoder. In low bit rate image compression, experiments show that our method significantly outperforms JPEG and JPEG2000 and can produce a better visual result with sharp edges, rich textures, and fewer artifacts.

Nowadays, there is a clear trend toward increasing the number of remote-sensing images acquired and their average size. This leads to the need to compress the images for storage, dissemination, and transfer over communication lines where lossy compression techniques are more popular. The images to be compressed or some of their components are often noisy. They must therefore be compressed taking into account the properties of the noise. Due to the noise filtering effect obtained during lossy compression of noisy images, an optimal operating point (OOP) may exist. The OOP is a parameter that controls the compression for which the quality of the compressed image is closer (closest) to the corresponding noise-free image than the quality of the noisy (original, uncompressed) image according to some quantitative criterion (metric). In practice, it is important to know whether the OOP exists for a given image, because if the OOP exists, it is appropriate to perform the compression in the OOP or at least in its neighborhood. Since the real image is absent in practice, it is impossible to determine a priori whether the OOP exists or not. Here, we focus on three-channel-remote-sensing images and show that it is possible to easily predict the existence of the OOP. Furthermore, it is possible to predict the metric values or their improvements with appropriate accuracy for practical use. The BPG (better portable graphics) encoder is considered a special case of an efficient compression technique. As an initial design step, the case of additive white Gaussian noise with equal variance in the three components is considered. While previous research was mainly focused on predicting the improvement (reduction) of the PSNR and PSNR-HVS-M metrics, here we focus on the modern visual quality metrics, namely PSNR-HA and MDSI. We also discuss what to do if, according to the prediction, an OOP is absent. Examples of lossy compression of noisy three-channel remote sensing images are given. It is also shown that the use of three-dimensional compression provides a compression ratio increase by several times compared with component-wise compression in the OOP.

The predictive lossy compression paradigm, which is emerging as an interesting alternative to conventional transform coding techniques, is studied. We first discuss this paradigm and outline the advantages and drawbacks with respect to transform coding. Next, we consider two low-complexity predictors and compare them under equal conditions on a large set of multispectral and hyperspectral images. Besides their rate-distortion performance, we attempt to gain some insight on the “quality” of the prediction residuals, comparing bit-rate and variance, and calculating the kurtosis. The results allow us to outline the directions for improvement of the algorithms, mainly in the treatment of noisy channels and the use of appropriate statistical models for the entropy-coding stage.