Power Signal-to-noise Ratio Research Articles

Realization of a high-efficient GDI-based 4:2 compressor for sum of absolute difference detection in image and signal processing

A new 4:2 approximate compressor is designed based on gate diffusion input (GDI), which has 14 transistors and only 6 errors in its outputs. The integration of the GDI with the dynamic threshold (DT) overcomes the body effect by the selection of carbon nanotube field-effect transistors (CNTFET) technology. By embedding the proposed compressor in the 8-bit sum of absolute difference (SAD) system, an area of 284.6 µm2 is occupied and a reliable system for difference detection between two images and electrocardiogram (ECG) results. Considering the fair conditions of comparison, the proposed circuit has significant performance in terms of circuitry parameters such as power, delay, and figure of merits (FoMs). In terms of power-delay-product (PDP) and power-delay-area-product (PDAP) the proposed cell shows a saving of 27 % and 47 %, respectively. A 40 % improvement in terms of a FoM containing PDP and power signal-to-noise ratio (PSNR) compared to the closest competitor qualifies the proposed circuit over the other counterparts. The proposed cell has a high performance under the difference detection in images and ECGs.

High-efficient and error-resilient gate diffusion input-based approximate full adders for complex multistage rapid structures

Two new approximate full adders (FAs) are proposed by the multiplexers (MUXs) and OR gates with the gate diffusion input (GDI) technique. The cells are named GDI-based MUX approximate FAs (GMAFAs), GMAFA1, and GMAFA2. The threshold voltage drop of the GDI-based circuits is solved by carbon nanotube field-effect transistors (CNTFETs) technology and the dynamic threshold (DT) technique. The FAs have accurate Cout, approximate Sum, and a low number of transistors. They are low-power, high-speed, and energy-efficient for ripple-carry adders (RCAs). The GMAFA1 and GMAFA2 have a total area of 0.068 µm2 and 0.119 µm2, respectively, which demonstrate 27.8% and 54.55% improvement in power-delay-area-product (PDAP) in comparison with the best references. The results of error distance (ED), normalized mean error distance (NMED), power signal-to-noise ratio (PSNR), and energy-saving confirm the accuracy of the presented cells for image processing applications.

An Adaptive Richardson-Lucy Algorithm for Medical Image Restoration

Image restoration is the process of estimating the original image content from a degraded picture. In this paper, the Richardson-Lucy iterative algorithm was developed to improve the quality of degraded medical images. It has been assumed that medical images are exposed to two types of degradation. The first type is the blur function in the Gaussian form with different widths, i.e. σ = 1 , 2, and 3. The second type of degradation was assumed to be of the independent white Gaussian noise type with different signal-to-noise ratio values: SNR = 10, 50 , and 100. The results obtained from the adaptive filter are compared, quantitatively, with different conventional filters: inverse, Wiener, and constraint least square, by applying different measures, such as: power signal to noise ratio (PSNR), structural similarity index (SSID), and root mean square error (RMSE). The comparison showed that the adaptive recovery filter achieves better results.

Structure-preserving Contrast Enhancement of Fundus Images using Dualistic Sub-Image Bi-histogram Bezier Curve

Diabetic retinopathy is a common eye disease among diabetic patients which is caused by excessive sugar in the blood vessels that damage the retina. Fundus images are retina images that are captured and diagnosed by ophthalmologists. Ophthalmologists diagnose the progressive stages of diabetic retinopathy so that early detection of pre-diabetic retinopathy can be carried out. However, the quality of the fundus image can be associated with the brightness of the background and the indistinctive vessel contrast. This paper presents a novel extension of Bi-histogram Bezier curve contrast enhancement (BBCCE) based on the mean partition of its histogram. The disadvantage of having mean as the threshold partition is that the histogram distribution can be skewed due to an outlier. The proposed Dualistic Sub-Image Bi-histogram Bezier Curve Contrast Enhancement (DSI-BBCCE) method partitions the original histogram into two, using the median of the active dynamic intensity range of the input image and process two Bezier transform curves separately to replace the original cumulative density function curve as the median is not affected by the outlier. This DSI-BBCCE has the advantage of preserving the structure, median brightness and preventing over enhancement. The result shows that DSI-BBCCE has achieved a power signal to noise ratio (PSNR) of 20.08±0.94 dB, absolute mean brightness error (AMBE) of 20.15±1.89, structural similarity index model (SSIM) of 0.8096±0.0185, structure measure operator (SMO) of 3.2±1.10 and lightness measure order (LMO) of 200.90±44.19.

Denoising and fuel spray droplet detection from light-scattered images using deep learning

A deep learning-based method for denoising and detecting the gas turbine engine spray droplets in the light-scattered image (Mie scattering) is proposed for the first time. A modified U-Net architecture is employed in the proposed method to denoise and regenerate the droplets. We have compared and validated the performance of the modified U-Net architecture with standard conventional neural networks (CNN) and modified ResNet architectures for denoising spray images from the Mie scattering experiment. The modified U-Net architecture performed better than the other two networks with significantly lower Mean Squared Error (MSE) on the validation dataset. The modified U-Net architecture also produced images with the highest Power Signal to Noise Ratio (PSNR) compared to the other two networks. This superior performance of the modified U-Net architecture is attributed to the encoder-decoder structure. During downsampling, as part of the encoder, only the most prominent features of the image are selectively retained by excluding any noise. This reconstruction of the noise-free features has produced a more accurate and better denoised image. The denoised images are then passed through a center predictor CNN to determine the location of the droplets with an average error of 1.4 pixels. The trained deep learning method for denoising and droplet center detection takes about 2.13 s on a single graphics processing unit (GPU). This study shows the promise for real-time processing of the experimental data using the well-optimized network.

Comparative Study on Performance of Discrete Wavelength Transform and Huffman Compression Technique on 2D Signal

Nowadays, the development of technology which involves multimedia data is widely used to help better understanding in spreading information. Image is known as 2D signal which contain huge data especially a high resolution image. This paper shows the comparison of applying lossy and lossless compression on the image data. Image compression is necessary in reducing the size of image for storage or transmission purpose to support most of the application nowadays. The technique applied in this paper is the hybrid of Discrete Wavelet Transform (DWT) technique and Huffman coding technique which are classified as lossy and lossless compression, respectively. The performance of image compression are evaluated in terms of compression ratio, Mean Square Error (MSE), Power Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM) and computing time. Several types of evaluation can determine better technique to apply on specific type of application. The stand-alone of each DWT and Huffman technique are evaluated before applying hybrid of DWT and Huffman technique. After conducting a comprehensive observation, the hybrid technique can compress with ratio about 1:17 to 1:27 due to the support from DWT technique that apply filter concept. The MSE value is high with the average about 69 which contributes to low PSNR value with about 29 to 30 dB due to the relation of PSNR equation with MSE value. Besides, the SSIM value is 0.6 or about 40% far from the original image that affect the output image. Despite of that, the computing time is fast with about 3 to 4 seconds which has been improved from stand-alone Huffman technique. Therefore, hybrid compression is capable of supporting each other techniques in stand-alone technique.

Image Restoration Using Lucy Richardson Algorithm for Deblurring Images with Improved PSNR, SSIM, NC in Comparison with Wiener Filter

Aim: Image is the most powerful tool to analyze the information. Sometimes the captured image gets affected with blur and noise in the environment, which degrades the quality of the image. Image restoration is a technique in image processing where the degraded image can be restored or recovered to its nearest original image. Materials and Methods: In this research Lucy-Richardson algorithm is used for restoring blurred and noisy images using MATLAB software. And the proposed work is compared with Wiener filter, and the sample size for each group is 30. Results: The performance was compared based on three parameters, Power Signal to Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), Normalized Correlation (NC). High values of PSNR, SSIM and NC indicate the better performance of restoration algorithms. Lucy-Richardson provides a mean PSNR of 10.4086db, mean SSIM of 0.4173%, and NC of 0.7433% and Wiener filter provides a mean PSNR of 6.3979db, SSIM of 0.3016%, NC of 0.3276%. Conclusion: Based on the experimental results and statistical analysis using independent sample T test, image restoration using Lucy-Richardson algorithm significantly performs better than Wiener filter on restoring the degraded image with PSNR (P<0.001) and SSIM (P<0.001).

Which Method Is Providing Correct Geometry of Pancreas in Filtering and Segmenting Ultrasound Images?

Pancreas is a small organ approximately six inches long, located at upper abdomen and adjacent to the small intestine. It is located at the back and deep into the human body. Therefore, it difficult to obtain a pancreas image clearly using ultrasound machine. In order to solve this problem, MATLAB software was used to filter and segment the ultrasound pancreas images using various methods. The images were selected based on high quality images produced from five subjects using AplioMX device from Toshiba ultrasound machine with a 3.5MHz convex transducer. In this study, there were various technique of filter used including Kaun filter, Wiener filter, Frost filter and Anisotropic Diffusion filter in order to reduce spackle noise of ultrasound pancreas images. Then, the filtered images were measured using Mean Square Error (MSE), Power Signal to Noise Ratio (PSNR), Average Difference (AD), Normalized Cross-Correlation (NCC), Maximum Difference (MD), Structural Content (SC), and Normalized Absolute Error (NAE) formula to evaluate the best quality images before undergo the segmentation process. As the result, Wiener filter was selected. In the segmentation process, the active countor method and level set method were evaluated. Then, the area of binary image and error percentage were calculated. As a conclusion, it shows that the filtering process using Wiener filter and segmentation method using level sets method has been successfully done to produce the best geometry of pancreas image.

Large aperture static interferential multispectral image compression with motion compensation and region of interest coding

To perform better on interferential multispectral image compression, wavelet transform based on motion compensation and region of interest coding and asymmetric zerotree was proposed. According to the movement in the images and the characters of spectrum, the motion compensation and raising coefficients of intensive interference area were used in this wavelet. In this way, the relativity among the spectra is reduced and the intensive interference area coefficients can be coded preferentially. The asymmetric zerotree was proposed due to the unsymmetrical structure of images formed by the wavelet. Experimental results show that with eightfold compression, this algorithm outperforms the three-dimensional set partitioning in hierarchical trees (3DSPIHT) in the range of 0.35–0.49 dB in power signal-to-noise ratio (PSNR) and outperforms the 2DSPIHT in the range of 2.24–2.62 dB in PSNR. Compared to joint picture expert group 2000 (JPEG2000) upscaling method, the effect of intensive interference area is also improved 0.29–0.38 dB in PSNR. Meanwhile, the average of mean square error of spectra within 46 spectrum bands is reduced to 9.69.