Algorithm For Optimal Quantization Research Articles

Rate-accuracy optimized quantization algorithm based on ROI image coding in power line inspection

Rate-accuracy optimized quantization algorithm based on ROI image coding in power line inspection

Resource-Restricted Environments Based Memory-Efficient Compressed Convolutional Neural Network Model for Image-Level Object Classification

In the past decade, Convolutional Neural Networks (CNNs) have achieved tremendous success in solving complex classification problems. CNN architectures require an excessive number of computations to achieve high accuracy. However, these models are deficient due to the heavy cost of storage and energy, which prohibits the application of CNNs to resource-constrained edge-devices. Hence, developing aggressive optimization schemes for efficient deployment of CNNs on edge devices has become the most important requirement. To find the optimal approach, we present a resource-limited environment based memory-efficient network compression model for image-level object classification. The main aim is to compress CNN architecture by achieving low computational cost and memory requirements without dropping system’s accuracy. To achieve the said goal, we propose a network compression strategy, that works in a collaborative manner, where Soft Filter Pruning is first applied to reduce the computational cost of the model. In the next step, the model is divided into No-Pruning Layers (NP-Layers) and Pruning Layers (P-Layers). Incremental Quantization is applied to P-Layers due to irregular weights distribution, while for NP-Layers, we propose a novel Optimized Quantization algorithm for the quantization of weights up to optimal levels obtained from the Optimizer. This scheme is designed to achieve the best trade-off between compression ratio and accuracy of the model. Our proposed system is validated for image-level object classification on LeNet-5, CIFAR-quick, and VGG-16 networks using MNIST, CIFAR-10, and ImageNet ILSVRC2012 datasets respectively. We have achieved high compression ratio with negligible accuracy drop, outperforming the state-of-the-art methods.

Automatic Change Analysis in Satellite Images Using Binary Descriptors and Lloyd–Max Quantization

In this letter, we present a novel technique for unsupervised change analysis that leads to a method of ranking the changes that occur between two satellite images acquired at different moments of time. The proposed change analysis is based on binary descriptors and uses the Hamming distance as a similarity metric. In order to render a completely unsupervised solution, the obtained distances are further classified using vector quantization methods (i.e., Lloyd's algorithm for optimal quantization). The ultimate goal in the change analysis chain is to build change intensity maps that provide an overview of the severeness of changes in the area under analysis. In addition, the proposed analysis technique can be easily adapted for change detection by selecting only two levels for quantization. This discriminative method (i.e., between changed/unchanged zones) is compared with other previously developed techniques that use principal component analysis or Bayes theory as starting points for their analysis. The experiments are carried on Landsat images at a 30-m spatial resolution, covering an area of approximately 59×51 km <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> over the surroundings of Bucharest, Romania, and containing multispectral information.

The rate-distortion optimized quantization algorithm in Compressive Sensing

Rate-distortion optimization greatly improves performance of compression coding system. In this paper, the rate-distortion optimized quantization algorithm is proposed for block-based Compressive Sensing. Compressive Sensing is the emerging technology which can encode a signal into a small number of incoherent linear measurements and reconstruct the entire signal from relatively few measurements. In the algorithm the sampling measurements are quantized optimally based on the rate-distortion theory. For the coefficients near dead-zone the quantization level with best rate-distortion performance is chosen. Moreover, in order to acquire the best performance, a fast Lagrange multiplier solving method is proposed to find the optimal slope λ* of the rate-distortion curve at the given bit budget. Experimental results show that the proposed algorithm improves objective and subjective performances substantially. The average gain about 0.7dB can be achieved with the same rate.

Near lossless compression of hyperspectral images based on distributed source coding

Effective compression technique of on-board hyperspectral images has been an active topic in the field of hyperspectral remote sensintg. In order to solve the effective compression of on-board hyperspectral images, a new distributed near lossless compression algorithm based on multilevel coset codes is proposed. Due to the diverse importance of each band, a new adaptive rate allocation algorithm is proposed, which allocates rational rate for each band according to the size of weight factor defined for hyperspectral images subject to the target rate constraints. Multiband prediction is introduced for Slepian-Wolf lossless coding and an optimal quantization algorithm is presented under the correct reconstruction of Slepian-Wolf decoder, which minimizes the distortion of reconstructed hyperspectral images under the target rate. Then Slepian-Wolf encoder exploits the correlation of the quantized values to generate the final bit streams. Experimental results show that the proposed algorithm has both higher compression efficiency and lower encoder complexity than several existing classical algorithms.

Error analysis of the optimal quantization algorithm for obstacle problems

In the paper Bally and Pagès (2000) an algorithm based on an optimal discrete quantization tree is designed to compute the solution of multi-dimensional obstacle problems for homogeneous R d -valued Markov chains ( X k ) 0⩽ k⩽ n . This tree is made up with the (optimal) quantization grids of every X k . Then a dynamic programming formula is naturally designed on it. The pricing of multi-asset American style vanilla options is a typical example of such problems. The first part of this paper is devoted to the analysis of the L p -error induced by the quantization procedure. A second part deals with the analysis of the statistical error induced by the Monte Carlo estimation of the transition weights of the quantization tree.

Wavelet Image Compressor—MinImage

Nowadays, still images are used everywhere in the digital world. The shortages of storage capacity and transmission bandwidth make efficient compression solutions essential. A revolutionary mathematics tool, wavelet transform, has already shown its power in image processing. MinImage, the major topic of this paper, is an application that compresses still images by wavelets. MinImage is used to compress grayscale images and true color images. It implements the wavelet transform to code standard BMP image files to LET wavelet image files, which is defined in MinImage. The code is written in C++ on the Microsoft Windows NT platform. This paper illustrates the design and implementation details in Min-Image according to the image compression stages. First, the preprocessor generates the wavelet transform blocks. Second, the basic wavelet decomposition is applied to transform the image data to the wavelet coefficients. The discrete wavelet transforms are the kernel component of MinImage and are discussed in detail. The different wavelet transforms can be plugged in to extend the functionality of MinImage. The third step is the quantization. The standard scalar quantization algorithm and the optimized quantization algorithm, as well as the dequantization, are described. The last part of MinImage is the entropy-coding schema. The reordering of the coefficients based on the Peano Curve and the different entropy coding methods are discussed. This paper also gives the specification of the wavelet compression parameters adjusted by the end user. The interface, parameter specification, and analysis of MinImage are shown in the final appendix.

Optimal Quantization of the Rayleigh Probability Distribution

We have derived an efficient and general algorithm for optimal quantization and applied it to the important, but neglected Rayleigh random variable. The optimal quantization range and output levels for the distortion criterion of minimum mean squared error have been calculated and tabulated for numbers of output levels from 1 to 64. We also compute and report the corresponding distortions and output entropies. For the sake of comparison, we also calculate and tabulate level spacings, minimum errors, and entropies for the optimal uniform (equal step size) quantization of the Rayleigh density. Distortion-rate curves are presented for uncoded and entropy coded quantizer outputs and compared with the distortion-rate function. We judge that these results will be useful to workers in the fields of optical holography, image processing, and communications.