scholarly journals Compressive Sensing for Natural Images

10.29007/xjzx ◽  
2018 ◽  
Author(s):  
Mayur Sevak ◽  
Anish Bagga ◽  
Arpita Agarwal ◽  
Krusha Jani

The Compressive sensing technique is a new era of arising platform for signal processing and data acquisition. The significant statement of Compressive sensing is that recovery of certain images or signals from fewer samples than required. On the encoding side two properties of a signal is required that are incoherence and sparsity. Initially, the signal is converted into specific transform i.e. wavelet using sensing matrix it takes required coefficients that has less dimensionality than the image dimensions and thereby, we get resultant matrix which is also called as measurements which in turn are non-adaptive. Similarly, on the decoding side, due to low dimension of transmitted vector matrix, convex optimization is required to solve this problem. Convex optimization (L1 minimization) provides an answer to undetermined linear systems without the knowledge of nature of undergoing parameters through the systems.In this paper, the work is being done on natural image compression. Compression of various black and white natural images is being done with help of ‘haar wavelet’ with second level decomposition and also reconstruct the same with three measurements of 60%, 70% and 80%. Also we have measured same with PSNR (Peak Signal to Noise Ratio), RMSE (Root Mean Square Error), CoC (Correlation Coefficients) of them.

Perception ◽  
1996 ◽  
Vol 25 (1_suppl) ◽  
pp. 85-85
Author(s):  
G M Kennedy ◽  
D J Tolhurst

Previous studies with simplified stimuli such as combinations of sinusoidal gratings have revealed phase identification losses in the periphery that are not eliminated by a scaling factor. How do these phase processing problems influence our ability to discriminate natural images in the periphery? In this study the ability of an observer to identify the ‘odd-image-out’ when there is either an amplitude-only, phase-only, or amplitude and phase change in one out of three stimuli is compared. Pairs of Fourier-manipulated black-and-white digitised photographs of natural images were used and phase and amplitude spectral exchanges of varying proportions were made between two different images. Measurements were made to determine the smallest phase change needed in order for the observer to reliably discriminate the manipulated image, compared to two reference stimuli, at eccentricities of 0°, 2.5°, 5°, and 10°. This was compared to discrimination thresholds found when amplitude and phase, and amplitude alone were exchanged. The ability to discriminate images on the basis of phase information alone did fall off quickly with eccentricity (comparable to phase and amplitude discriminations). However, there was a much more rapid decline in amplitude-only discrimination. It appears that phase information in natural scenes remains a relatively more important visual cue in the periphery than amplitude.


2018 ◽  
Author(s):  
Arvind Iyer ◽  
Johannes Burge

AbstractTo model the responses of neurons in the early visual system, at least three basic components are required: a receptive field, a normalization term, and a specification of encoding noise. Here, we examine how the receptive field, the normalization factor, and the encoding noise impact the model neuron responses to natural images and the signal-to-noise ratio for natural image discrimination. We show that when these components are modeled appropriately, the model neuron responses to natural stimuli are Gaussian distributed, scale-invariant, and very nearly maximize the signal-to-noise ratio for stimulus discrimination. We discuss the statistical models of natural stimuli that can account for these response statistics, and we show how some commonly used modeling practices may distort these results. Finally, we show that normalization can equalize important properties of neural response across different stimulus types. Specifically, narrowband (stimulus- and feature-specific) normalization causes model neurons to yield Gaussian-distributed responses to natural stimuli, 1/f noise stimuli, and white noise stimuli. The current work makes recommendations for best practices and it lays a foundation, grounded in the response statistics to natural stimuli, upon which principled models of more complex visual tasks can be built.


Author(s):  
Xingsong Hou ◽  
Lan Zhang ◽  
Zan Chen ◽  
Chen Gong

Compressive sensing (CS) has been proposed for images that are sparse under a certain transform domain. However, many natural images are not strictly sparse in the transform domain, causing a tail-folding effect that degrades the performance of the CS reconstruction. To decrease such effect, a sparse-filtering (SF) in Directional Lifting Wavelet Transform (DLWT) domain based Bayesian compressive sensing (BCS) algorithm (DLWT-SF-TSW-BCS) is proposed. At the encoder, DLWT, an efficient multi-scale geometrical analysis (MGA) tool, is used to produce the sparse representation for natural images. Then sparse-filtering is adopted to cut off the small DLWT coefficients before random measurement. At the decoder, the interscale tree-structure redundancy in DLWT domain is further exploited in Bayesian reconstruction. Experimental results show that the proposed DLWT-SF-TSW-BCS algorithm significantly outperforms other state-of-the-art CS reconstruction algorithms, for example, peak signal to noise ratio (PSNR) gain up to 10.00 dB over the tree structured wavelet compressive sensing (TSW-CS).


2013 ◽  
Vol 462-463 ◽  
pp. 391-398 ◽  
Author(s):  
Tian Yi Li ◽  
Pei Gen Yuan ◽  
Dong Yan Yang ◽  
Xi Yao Hu

We proposed an approach for estimating the density of salt-pepper noise in images with correlation inspection. The correlation coefficients histogram was introduced in this paper to depict the correlation distributions of images. Based on the fact that the correlation distributions of natural images are nearly independent of individual images, we revealed how the correlation coefficients histogram of the noisy image deviates from that of natural image along with the noise density in qua- ntitative form, thus we took advantage of this relation to make estimation. Simulation results showed that the proposed approach outperforms those of existing methods.


2013 ◽  
Vol 11 (1) ◽  
pp. 8-13
Author(s):  
V. Behar ◽  
V. Bogdanova

Abstract In this paper the use of a set of nonlinear edge-preserving filters is proposed as a pre-processing stage with the purpose to improve the quality of hyperspectral images before object detection. The capability of each nonlinear filter to improve images, corrupted by spatially and spectrally correlated Gaussian noise, is evaluated in terms of the average Improvement factor in the Peak Signal to Noise Ratio (IPSNR), estimated at the filter output. The simulation results demonstrate that this pre-processing procedure is efficient only in case the spatial and spectral correlation coefficients of noise do not exceed the value of 0.6


Author(s):  
IRMA SAFITRI ◽  
NUR IBRAHIM ◽  
HERLAMBANG YOGASWARA

ABSTRAKPenelitian ini mengembangkan teknik Compressive Sensing (CS) untuk audio watermarking dengan metode Lifting Wavelet Transform (LWT) dan Quantization Index Modulation (QIM). LWT adalah salah satu teknik mendekomposisi sinyal menjadi 2 sub-band, yaitu sub-band low dan high. QIM adalah suatu metode yang efisien secara komputasi atau perhitungan watermarking dengan menggunakan informasi tambahan. Audio watermarking dilakukan menggunakan file audio dengan format *.wav berdurasi 10 detik dan menggunakan 4 genre musik, yaitu pop, classic, rock, dan metal. Watermark yang disisipkan berupa citra hitam putih dengan format *.bmp yang masing-masing berukuran 32x32 dan 64x64 pixel. Pengujian dilakukan dengan mengukur nilai SNR, ODG, BER, dan PSNR. Audio yang telah disisipkan watermark, diuji ketahanannya dengan diberikan 7 macam serangan berupa LPF, BPF, HPF, MP3 compression, noise, dan echo. Penelitian ini memiliki hasil optimal dengan nilai SNR 85,32 dB, ODG -8,34x10-11, BER 0, dan PSNR ∞.Kata kunci: Audio watermarking, QIM, LWT, Compressive Sensing. ABSTRACTThis research developed Compressive Sensing (CS) technique for audio watermarking using Wavelet Transform (LWT) and Quantization Index Modulation (QIM) methods. LWT is one technique to decompose the signal into 2 sub-bands, namely sub-band low and high. QIM is a computationally efficient method or watermarking calculation using additional information. Audio watermarking was done using audio files with *.wav format duration of 10 seconds and used 4 genres of music, namely pop, classic, rock, and metal. Watermark was inserted in the form of black and white image with *.bmp format each measuring 32x32 and 64x64 pixels. The test was done by measuring the value of SNR, ODG, BER, and PSNR. Audio that had been inserted watermark was tested its durability with given 7 kinds of attacks such as LPF, BPF, HPF, MP3 Compression, Noise, and Echo. This research had optimal result with SNR value of 85.32 dB, ODG value of -8.34x10-11, BER value of 0, and PSNR value of ∞.Keywords: Audio watermarking, QIM, LWT, Compressive Sensing.


Author(s):  
Martina Ladrova ◽  
Radek Martinek ◽  
Jan Nedoma ◽  
Marcel Fajkus

Electromyogram (EMG) recordings are often corrupted by the wide range of artifacts, which one of them is power line interference (PLI). The study focuses on some of the well-known signal processing approaches used to eliminate or attenuate PLI from EMG signal. The results are compared using signal-to-noise ratio (SNR), correlation coefficients and Bland-Altman analysis for each tested method: notch filter, adaptive noise canceller (ANC) and wavelet transform (WT). Thus, the power of the remaining noise and shape of the output signal are analysed. The results show that the ANC method gives the best output SNR and lowest shape distortion compared to the other methods.


2018 ◽  
Vol 56 ◽  
pp. 84-93 ◽  
Author(s):  
A. L. Shulyar

An important selection peculiarity of dairy cows is their productive longevity. Besides, the indicators of cow productivity along with genetic potential are the most important criteria of the economic efficiency of their use. Low or not sufficient level of lifetime productivity indicators and of the period of the economic use of cows determine the necessity of monitoring of the selection information according to these indicators in the breed herds as well as the of using corresponding selection methods for their improvement. The goal of our research was the estimation of an economic use of lifetime cow productivity of Ukrainian Black-and-White dairy breed depending on the reasons of their disposal from the herd and genotype. The research was conducted at a private agrarian firm (PAF) «Yerchyky» Zhytomyr oblast according to the data of breeding and productive use of 497 cows which were disposed from the herd on different reasons. The reasons for the cow disposal were determined according to the veterinary disposal acts. The indicators of cow economic use and lifetime productivity were determined on the ground of the term of rearing, life duration, term of economic use, amount of lactations, the total duration of lactation period, milk yields, milk yield per one lactation day, per life, and per economic use. Statistical processing was performed by means of Microsoft Excel software kit using the methods of variation statistics. As follows from the results of the research, in the process of holsteinization the dynamics of the parameters of the economic use of cows is characterized by curvilineal type: as follows from the heterotic effect they increased in animals with an conditional share of genetic heredity of Holstein breed from 50 and 62.5 to 75%, and then when the genetic heredity share of the improved breed increased by more than 75% they got worse due to the contradiction “genotype-environment”. At the same time within the determined genotypes (1st group-up to 50%, 5th group-87.6–100%) the indicators of lifetime productivity increase in case of 66% of reliable results. Besides, the interconnection level of the indices of lifetime use of the dairy cows of different genotypes has been studied. The criteria of the significance of the correlation coefficients according to Student has been used. The closer the linkage, the higher the correlation coefficients and the higher the criteria of their reliability. Generalized criteria of reliability on the grounds of all indicators of lifetime use was the lowest in cows of the 1st and 2nd groups, and the highest in cows of the 3rd group with further decrease in high half-bred cows and pure-breed cows of the 4th and 5th groups. It testifies to the fact that there is a close interconnection between given parameters and real milk yields in 3rd group cows (62.6–75.0% of the improved Holstein breed). At present, the reason of the shortening of the period of economic use is an early disposal of cows from the herd. A greater part of milking cows is disposed in the most productive period or even before it. At the given enterprise, a vast majority of holsteinized cows (77.5%) was disposed from the herd on four main reasons: unsatisfactory reproductive capability (32.2%), limbs diseases (22.3%), postpartal complications (13.9%) and low productivity (9.1%). The other reasons for cow disposal were within 1.6–7.0% (totaled up 22.5%) as to the total number of cow population that were disposed of the herd. It should be mentioned that the disposal caused by low productivity was only in the first calvers of the 1st and 2nd groups, in which the conditional share of genetic heredity of a Holstein breed did not exceed 62.5%. The parameters of the economic use and of lifetime productivity of Ukrainian Black-and-White dairy cows of different genotypes which were disposed of the herd because of their unsatisfactory reproductive capability, limbs diseases, postpartal complications and low milking productivity have been also analyzed. It has been determined that the indicated reasons for the disposal had different effects on the parameters of an economic use and on cow productivity. Because of any of four indicated reasons for cow disposal of the herd under condition of in-barn keeping and optimal feeding the preference on lifetime productivity belongs to the cows of the 4th and 5th groups (with a conditional share of Holstein heredity of more than 75%) with simultaneous worsening of the indicators of economic use. As a third part of cows at this enterprise was disposed of the herd ahead of time on the reason of unsatisfactory reproductive capability, that results in breeding value and in the results of the economic use of cows, the updating of the technology of artificial animal insemination and renewal will promote to solving the problem.


2017 ◽  
Vol 53 ◽  
pp. 139-148
Author(s):  
М. І. Kuziv ◽  
Е. І. Fedorovych ◽  
N. M. Kuziv ◽  
I. V. Novak

The productivity cow dairy breed largely depends on the careful evaluation and selection of animals by milk production and the type of exterior. The estimation of animals by measuring of body plays an important role in plant-breeding work in the process of further improvement of the newly formed Ukrainian dairy breeds by the type. It’s because exterior is one of the main signs of breeding dairy cattle together with indicators of milk production, reproductive ability and productive longevity. The aim of research was to investigate the formation of milk productivity of cows black and white dairy cattle based on their body measurements after the first calving. The research was done on animals Ukrainian black-and-white dairy breed in the «Milk River» farm in the Sokal and Brody offices breeding reproducers «Breeder» Lviv region and plant-breeding «Yamnytsya» Ivano-Frankivsk region. The body measurements of first calving cows (height at withers and depth of chest, breast width, chest girth, scythe body length, the width of the pelvic bones and metacarpus circumference cow) and milk productivity (the milk yield, fat content in milk and milk fat) cows first, second, third and better lactation were studied by the materials zoo technical accounting. The force of influence on performance metrics was calculated by Single-factor disperse analysis method. The results of research were treated by variational statistics by G.F. Lakin. Established that the milk yield of cows Ukrainian black-and-white dairy cattle depends on body measurements after the first calving. The cows Ukrainian black-and-white dairy breed in Western Ukraine that after the first calving had a height at the withers 134-136,9 cm, depth of chest – 76 cm or more, chest width – 47 cm or more, chest girth by shoulder blades – 196-203,9 cm, scythe body length – 160-164,9 cm, the width of the pelvic bones – 53 cm and metacarpus circumference cow –19-20,4 cm, are the most productive. Animals with these measurements of the body most of the largest number of yield milk and milk fat had probable advantage over cows with other indicators of these measurements. By the fat content in milk by the first, second, third and better lactation there was no significant difference between animals with different body measurements after the first calving in most cases. Among the studied parameters milk productivity of cows and measurements of their bodies after first calving we found varying strength and direction of correlation. The correlation coefficients between body measurements firstborns and yield milk by the first, second, third and better lactation were positive and depended on measurements and lactation ranged from +0,297 to +0,478. Between such measurements of firstborns, as the height at the withers, the depth of chest, width of chest, chest girth in blades, the width of the pelvic bones and with fat content in milk for the first, second, third and better lactation correlation coefficients were though low, but positive, and dependently on measurements and lactation were within +0,026 – +0,109. The correlation coefficients between the oblique body length after the first calving cows and fat content in milk were negative (-0,045 – -0,005) and between metacarpus circumference and fat ranged from negative to positive values (-0,038 – +0,017). The correlation coefficients between measurements body in firstborn and milk fat yield in the first, second, third and better lactation were positive and dependently on measurements and lactation ranged from +0,288 to +0,473. The likely positive correlation coefficient between soundings article body and yield milk and milk fat yield indicates that selection of cows by exterior will ensure the effectiveness of selection by milk productivity. The impact of body measurements cows after first calving on the milk yield value dependently on lactation yield was 7,6-22,9 %, on the yield of milk fat – 7,1-22,9 %. The greatest influence on these indicators were height at the withers, the depth of the chest, chest girth shoulder blades and scythe body length, somewhat lower impact had breast width and the width of the pelvic bones and lowest – metacarpus circumference. The impact of body measurements cows after first calving on the fat content in milk was negligible and, dependently on measurements and lactation ranged from 0,1 to 1,6 %.


Sign in / Sign up

Export Citation Format

Share Document