Inverse Discrete Wavelet Transform Research Articles

Pixel-Level Fusion Approach with Vision Transformer for Early Detection of Alzheimer’s Disease

Alzheimer’s disease (AD) has become a serious hazard to human health in recent years, and proper screening and diagnosis of AD remain a challenge. Multimodal neuroimaging input can help identify AD in the early mild cognitive impairment (EMCI) and late mild cognitive impairment (LMCI) stages from normal cognitive development using magnetic resonance imaging (MRI) and positron emission tomography (PET). MRI provides useful information on brain structural abnormalities, while PET data provide the difference between physiological and pathological changes in brain anatomy. The precision of diagnosing AD can increase when these data are combined. However, they are heterogeneous and appropriate, and an adequate number of features are required for AD classification. This paper proposed a multimodal fusion-based approach that uses a mathematical technique called discrete wavelet transform (DWT) to analyse the data, and the optimisation of this technique is achieved through transfer learning using a pre-trained neural network called VGG16. The final fused image is reconstructed using inverse discrete wavelet transform (IDWT). The fused images are classified using a pre-trained vision transformer. The evaluation of the benchmark Alzheimer’s disease neuroimaging initiative (ADNI) dataset shows an accuracy of 81.25% for AD/EMCI and AD/LMCI in MRI test data, as well as 93.75% for AD/EMCI and AD/LMCI in PET test data. The proposed model performed better than existing studies when tested on PET data with an accuracy of 93.75%.

D2Net: Deep Denoising Network in Frequency Domain for Hyperspectral Image

Dear Editor, Since the existing hyperspectral image denoising methods suffer from excessive or incomplete denoising, leading to information distortion and loss, this letter proposes a deep denoising network in the frequency domain, termed D2Net. Our motivation stems from the observation that images from different hyperspectral image (HSI) bands share the same structural and contextual features while the reflectance variations in the spectra are mainly fallen on the details and textures. We design the D2Net in three steps: 1) spatial decomposition, 2) spatial-spectral denoising, and 3) refined reconstruction. It achieves multi-scale feature learning without information loss by adopting the rigorous symmetric discrete wavelet transform (DWT) and inverse discrete wavelet transform (IDWT). In particular, the specific design for different frequency components ensures complete noise removal and preservation of fine details. Experiment results demonstrate that our D2Net can attain a promising denoising performance.

Intelligent diagnosis of flip chip solder joints with resolution enhanced SAM image

Scanning acoustic microscopy (SAM) technique has been applied to defect inspection in electronic devices. With the increase of packaging density, detection of the micro-defects in high density devices becomes more and more challenging. The SAM test is suffering from sacrificing the spatial resolution to reach a certain penetration depth of the ultrasonic waves. So it is necessary to enhance the resolution level of the SAM image. In this paper, a wavelet based resolution enhancement technique was investigated to reconstruct a high quality image for SAM test of the flip chip packages. The stationary wavelet transform was adopted to decompose the captured SAM image into four frequency subbands, and the high frequency subbands were enhanced by adding the difference matrix in the intermediate stage, and a super resolution SAM image was derived from combining all the subbands by using the Inverse Discrete Wavelet Transform. Then the solder joints segmented from the SR-SAM image were classified by using the SVM algorithm. The results validated that the proposed technique is effective to improve the detection accuracy of SAM test.

Pansharpening via Frequency-Aware Fusion Network With Explicit Similarity Constraints

The process of fusing a high spatial resolution (HR) panchromatic (PAN) image and a low spatial resolution (LR) multispectral (MS) image to obtain an HRMS image is known as pansharpening. With the development of convolutional neural networks, the performance of pansharpening methods has been improved, however, the blurry effects and the spectral distortion still exist in their fusion results due to the insufficiency in details learning and the frequency mismatch between MS and PAN. Therefore, the improvement of spatial details at the premise of reducing spectral distortion is still a challenge. In this paper, we propose a frequency-aware fusion network (FAFNet) together with a novel high-frequency feature similarity loss to address above mentioned problems. FAFNet is mainly composed of two kinds of blocks, where the frequency aware blocks aim to extract features in the frequency domain with the help of discrete wavelet transform (DWT) layers, and the frequency fusion blocks reconstruct and transform the features from frequency domain to spatial domain with the assistance of inverse DWT (IDWT) layers. Finally, the fusion results are obtained through a convolutional block. In order to learn the correspondence, we also propose a high-frequency feature similarity loss to constrain the HF features derived from PAN and MS branches, so that HF features of PAN can reasonably be used to supplement that of MS. Experimental results on three datasets at both reduced- and full-resolution demonstrate the superiority of the proposed method compared with several state-of-the-art pansharpening models. The codes are available at https://github.com/YinghuiXing/FAFNet.

Image Denoising using Wavelet Cycle Spinning and Non-local Means Filter

Removing as much noise as possible in an image while preserving its fine details is a complex and challenging task. We propose a wavelet-based and non-local means (NLM) denoising method to overcome the problem. Two well-known wavelets: dual-tree complex wavelet transform (DT-CWT) and discrete wavelet transform (DWT), have been used to change the noise image into several wavelet coefficients sequentially. NLM filtering and universal hard thresholding with cycle spinning have been used for thresholding on its approximation and detail coefficients, respectively. The inverse two-dimensional DWT was applied to the modified wavelet coefficients to obtain the denoised image. We conducted experiments with twelve test images on the set12 data set, adding the additive Gaussian white noise with variances of 10 to 90 in increments of 10. Three evaluation metrics, such as peak signal noise to rate (PSNR), structural similarity index metric (SSIM), and mean square error (MSE), have been used to evaluate the effectiveness of the proposed denoising method. From these measurement results, the proposed denoising method outperforms DT-CWT, DWT, and NLM almost in all noise levels except for the noise level of 10. At that noise level, the proposed denoising method is lower than NLM but better than DT-CWT and DWT.

Fusion of overexposed and underexposed images using caputo differential operator for resolution and texture based enhancement

The visual quality of images captured under sub-optimal lighting conditions, such as over and underexposure may benefit from improvement using fusion-based techniques. This paper presents the Caputo Differential Operator-based image fusion technique for image enhancement. To effect this enhancement, the proposed algorithm first decomposes the overexposed and underexposed images into horizontal and vertical sub-bands using Discrete Wavelet Transform (DWT). The horizontal and vertical sub-bands are then enhanced using Caputo Differential Operator (CDO) and fused by taking the average of the transformed horizontal and vertical fractional derivatives. This work introduces a fractional derivative-based edge and feature enhancement to be used in conjuction with DWT and inverse DWT (IDWT) operations. The proposed algorithm combines the salient features of overexposed and underexposed images and enhances the fused image effectively. We use the fractional derivative-based method because it restores the edge and texture information more efficiently than existing method. In addition, we have introduced a resolution enhancement operator to correct and balance the overexposed and underexposed images, together with the Caputo enhanced fused image we obtain an image with significantly deepened resolution. Finally, we introduce a novel texture enhancing and smoothing operation to yield the final image. We apply subjective and objective evaluations of the proposed algorithm in direct comparison with other existing image fusion methods. Our approach results in aesthetically subjective image enhancement, and objectively measured improvement metrics.

Robust Image Encryption Scheme for Biometric Finger Print Images using Bit Planes, DWT & Cubic Map

UIDAI employs biometric finger print system to generate omnipresent identity card for Indian citizens. Biometric finger print images are therefore accepted by all Banks and Government offices to established identity of the person. Furthermore lots of private organizations employ biometric finger print system to record attendance and personal information of its employees. Consequently these biometric finger print images are stored and transmitted widely enhancing risk of misusing user personal information and cybercrimes. In this paper, robust image encryption scheme for biometric finger print images is proposed through bit planes; cubic map and DWT. The complete encryption or decryption process can be decomposed into three methods. Firstly each 8-bit biometric image is separated into 8 bit planes corresponding to each bit. Secondly DWT is applied that further encrypts the image in frequency domain by manipulating LL subbands through cubic map. Finally inverse DWT (IDWT) converts each image into spatial domain which is synthesis to from a single 8-bit encrypted image. The robustness of the proposed encryption scheme is measured through entropy, correlation coefficient, number of pixel change rate (NPCR) and unified average change intensity (UACI). Furthermore the proposed encryption scheme is also tested successfully for compression and noise attacks.

MLWAN: Multi-Scale Learning Wavelet Attention Module Network for Image Super Resolution

Image super resolution (SR) is an important image processing technique in computer vision to improve the resolution of images and videos. In recent years, deep convolutional neural network (CNN) has made significant progress in the field of image SR; however, the existing CNN-based SR methods cannot fully search for background information in the measurement of feature extraction. In addition, in most cases, different scale factors of image SR are assumed to be different assignments and completed by training different models, which does not meet the actual application requirements. To solve these problems, we propose a multi-scale learning wavelet attention network (MLWAN) model for image SR. Specifically, the proposed model consists of three parts. In the first part, low-level features are extracted from the input image through two convolutional layers, and then a new channel-spatial attention mechanism (CSAM) block is concatenated. In the second part, CNN is used to predict the highest-level low-frequency wavelet coefficients, and the third part uses recursive neural networks (RNN) with different scales to predict the wavelet coefficients of the remaining subbands. In order to further achieve lightweight, an effective channel attention recurrent module (ECARM) is proposed to reduce network parameters. Finally, the inverse discrete wavelet transform (IDWT) is used to reconstruct HR image. Experimental results on public large-scale datasets demonstrate the superiority of the proposed model in terms of quantitative indicators and visual effects.

Deep learning-based classification of healthy aging controls, mild cognitive impairment and Alzheimer’s disease using fusion of MRI-PET imaging

Automated detection of dementia stage using multimodal imaging modalities will be helpful for improving the clinical diagnosis. In this study, we develop the Inception-ResNet wrapper model in differentiating the healthy controls (HC), mild cognitive impairment (MCI), and Alzheimer’s disease (AD) using conjoint magnetic resonance imaging (MRI) and positron emission tomography (PET) scans. We use T1-weighted MR and PET images of individuals aged between 42 and 95 years, including HC, MCI and AD patients. We first perform 3D tissue segmentation of MR images after skull striping. The atlas-based segmented MR image tissue is fused with PET image. Then we transform PET images from RGB to HSI color space and apply fusion of MRI with PET images using two-dimensional Fourier and discrete wavelet transform (DWT) and then reconstruct the MR-PET fused image using inverse Fourier and DWT methods. After the fusion of MRI and PET imaging modalities, we used 60 % training, 20 % for validation and the remaining 20 % as a test set using various convolutional neural networks. We found the proposed model as the best classifier with an accuracy of 95.5 %, 94.1 % and 95.9 % in classifying HC vs MCI, MCI vs AD and AD vs HC respectively when compared to the existing methods. We conclude that the proposed deep learning model has potential in automated classification of healthy and dementia stages using combined MRI and PET modalities with good performance.

Using Discrete Wavelet Transform and Wiener filter for Image De-nosing

In this paper we proposed an algorithm for image de-nosing based on; the two level discrete wavelet transform (DWT), and Wiener filter, also this paper describe estimate noise power. At first The DWT transform noisy image into sub-bands, consist of lowfrequencyand high-frequencies, and then estimate noise power for each sub-band. The noise power is computed through two important computations; compute square of variance for each sub-band then compute the mean of the variance. After compute the variance apply the wiener filter on each sub-band by using local window nn , finally perform inverse DWT to obtain de-noised image. Our algorithm tested on the two color images and also compared with Normal Shrink filter and Wiener filter.

Image Reconstruction with Multiscale Interest Points Based on a Conditional Generative Adversarial Network

A new image reconstruction (IR) algorithm from multiscale interest points in the discrete wavelet transform (DWT) domain was proposed based on a modified conditional generative adversarial network (CGAN). The proposed IR-DWT-CGAN model generally integrated a DWT module, an interest point extraction module, an inverse DWT module, and a CGAN. First, the image was transformed using the DWT to provide multi-resolution wavelet analysis. Then, the multiscale maxima points were treated as interest points and extracted in the DWT domain. The generator was a U-net structure to reconstruct the original image from a very coarse version of the image obtained from the inverse DWT of the interest points. The discriminator network was a fully convolutional network, which was used to distinguish the restored image from the real one. The experimental results on three public datasets showed that the proposed IR-DWT-CGAN model had an average increase of 2.9% in the mean structural similarity, an average decrease of 39.6% in the relative dimensionless global error in synthesis, and an average decrease of 48% in the root-mean-square error compared with several other state-of-the-art methods. Therefore, the proposed IR-DWT-CGAN model is feasible and effective for image reconstruction with multiscale interest points.

DnSwin: Toward real-world denoising via a continuous Wavelet Sliding Transformer

Real-world image denoising is a practical image restoration problem that aims to obtain clean images from in-the-wild noisy inputs. Recently, the Vision Transformer (ViT) has exhibited a strong ability to capture long-range dependencies, and many researchers have attempted to apply the ViT to image denoising tasks. However, a real-world image is an isolated frame that makes the ViT build long-range dependencies based on the internal patches, which divides images into patches, disarranges noise patterns and damages gradient continuity. In this article, we propose to resolve this issue by using a continuous Wavelet Sliding-Transformer that builds frequency correspondences under real-world scenes, called DnSwin. Specifically, we first extract the bottom features from noisy input images by using a convolutional neural network (CNN) encoder. The key to DnSwin is to extract high-frequency and low-frequency information from the observed features and build frequency dependencies. To this end, we propose a Wavelet Sliding-Window Transformer (WSWT) that utilizes the discrete wavelet transform (DWT), self-attention and the inverse DWT (IDWT) to extract deep features. Finally, we reconstruct the deep features into denoised images using a CNN decoder. Both quantitative and qualitative evaluations conducted on real-world denoising benchmarks demonstrate that the proposed DnSwin performs favorably against the state-of-the-art methods.

Wavelet descriptor network for arbitrary-shaped text detection

Arbitrary-shaped scene text detection is still a challenging task in computer vision, due to its high versatility and complexity. Most methods directly describe text instances by rectangles, quadrangles, or curves, probably leading to insufficient representations. We introduce wavelet descriptor (WD) for arbitrary-shaped text representation that decomposes text contours to wavelet coefficients and extracts feature sequences by discrete wavelet transform (DWT). Based on WD, we propose WD network, which predicts wavelet coefficients then reconstructs text instances accurately and smoothly via inverse DWT and nonmaximum suppression. We further study the restoration capability of various wavelet functions for text contour, including sym5, bior3.1, bior4.4, coif3, db5, dmey, and rbio2.8. To evaluate the proposed method, we conduct experiments on available public datasets. The F-measure of our method achieves 85.8% on challenging curved dataset CTW1500 and higher than 85% on Total-Text and ICDAR2015, demonstrating the excellent performance of our method.

A color channel multiplexing approach for robust discrete wavelet transform based image watermarking

SummaryCopyright protection of digital images has become a consequential issue in recent years. To prevent a breach of copyright for a color image, this article presents a novel discrete wavelet transform (DWT) based robust image watermarking method with a color channel multiplexing approach. In this proposed method, the original color image is separated into red, green, and blue color channels. One of these color channels is copied and a set of four channels is obtained. Then these channels are combined to obtain a new gray‐level cover image. 2‐level DWT is performed on the cover image. Arnold scrambled binary watermark image i embedded into the low‐frequency sub‐band of the cover image. 2‐level inverse DWT is performed to produce a gray‐level watermarked image. Then a demultiplexing process is performed by separating the gray‐level watermarked image into four color channels. The average values of the two copied color channels are calculated and then converted into a single‐color channel. Finally, the watermarked color image is get by merging three different color channels. To approve the performance of the presented method, various experimental tests have been performed and a benchmarking analysis has been conducted. The results show that the proposed method is robust against non‐linear and linear attacks while protecting watermarking imperceptibility.

DWT based Image Super Resolution using Segmentation and Edge Detection

An image super-resolution is the development of creating a high-resolution image from low-resolution images for better image visualization for several recent applications. In this research paper, we propose DWT-based Image Super Resolution using Segmentation and Edge Detection for better picturing of images. The different kinds of colore images are considered and converted into greyscale images and resized to a uniform size of 166x304. The Discrete Wavelet Transform (DWT) is used to get one low-frequency and three high-frequency bands with a size of 83x152. The four bands are fused and dividing each coefficient by two to get Low Resolution (LR) images. Each coefficient of the LR image is inflated into 2x2 coefficients to get High Resolution (HR) image. The HR matrix is segmented into a 3x3 overlapped matrix and an average value is calculated and assigned to the whole HR matrix. The Canny edge detection is used on the original image to get the edge detected image. The pixel values of HR and canny edge detected images are compared and high-value coefficients are considered. The guided filter is applied to the highvalue coefficients matrix to improve the quality of the image by refining the image. The LL band of the DWT matrix is padded with zeros to convert matrix size to 166x304 and Inverse Discrete Wavelet Transform (IDWT) is used on it. The refined and IDWT images are merged by computing the average values to acquire Super Resolution (SR) image. It is noted that the anticipated system results are enhanced compared to current methods. Keywords: DWT, Fusion, Image Processing, Segmentation, Super Resolution

An asymmetric image encryption scheme based on hash SHA-3, RSA and compressive sensing

ObjectiveIn order to avoid the unauthorized access to private images and make a secure communication, an asymmetric image encryption algorithm based on SHA-3 and compressive sensing is proposed. MethodsFirstly, a random matrix is generated and a modular-addition operation is performed between it with the plain image, obtaining a preprocessed image. Then, hash values of the preprocessed image are computed by secure hash algorithm in the third generation (SHA-3), and are grouped and summed to get three plaintext keys. Using Rivest-Shamir-Adleman (RSA), three ciphertext keys can be obtained correspondingly. Secondly, a new mathematical transformation model (MTM) is designed to transform all keys into initial values for chaotic system. Then, keystream is calculated accordingly. Thirdly, the plain image is compressed by compressive sensing (CS), and further confused by random sequences. Then, applying discrete wavelet transformation (DWT) to the confused image and generating four components of high and low frequencies. Chaotic sequences are used again to confuse the low frequency components, and then all components are recombined again into a matrix. Thereafter, perform inverse DWT (IDWT) to get a middle cipher image (MCI). Finally, another random matrix is generated by chaotic sequences, and the final cipher image is obtained by a modular-addition operation again to the MCI. ResultsExperiments show that the proposed scheme can reconstruct the plain image with a high quality. In particular, the proposed method can resist known plaintext attack and chosen plaintext attack because the keystreams are related to the plain image and with no extra transmission.

Circulate Matrix and Compression Sensing Based Multi-Level Image Encryption

Digital data security is a broad research area in the field of science and technology. A lot of research was focused on information security-based mechanism for secure communication. This paper presents a novel image encryption as well as compression based on measurement matrix, pixel exchange and logistic cat map, which includes the permutation, compression, and diffusion processes. Initially the image is divided into four equal sizes of blocks and then each block is transformed into horizontal and vertical low and high frequency band. Then a random matrix multiplication function is applied to achieve an encrypted and scrambling frequency component and apply inverse DWT procedure to get first level of scrambled blocks, and further we apply the second level of security mechanism. Here each adjacent block pixel is exchanged by using the random matrices. For providing the high level of compression we design measurement matrices in compressive sensing by utilizing the circulate matrices and controlling the original column vectors of the circulate matrices with Arnold cat map. With the help of measurement matrix again the blocks are encrypted. Experimental results and performance analyses validate the good compression performance and high security of the given algorithm.

Image Fusion by Using Bi-Orthogonal Wavelets in Discrete Wavelet Transform (DWT) Domain

Image fusion can be widely explained as merging more than one photograph or a few in their factors right into a single photo without the creation of distortion or lack of information. Image fusion pastimes mix complementary as properly as approximate statistics for greater than one photograph to create fused photo output. Hence, the output picture achieved must have to include an extra accurate representation of the position than any of the man or woman supply pix and is greater fantastic for human visible, and pc understanding or in addition picture processing and comparison duties. . The notably used fusion rule is the most willpower scheme. This handy scheme chooses the biggest absolute wavelet coefficient at each location from the given snapshots because of the coefficient on the area within the fused photo. After that, the fused photo is obtained with the aid of using the inverse DWT for the corresponding wavelet coefficient. The well-known significantly used pixel-primarily based fusion rule is the aforementioned maximum favored scheme. This approach can choose the salient elements from the furnished photographs; however, it's miles touchy with noise and artifacts as they have been supposed to have better contrast. When the 2 photographs are degraded with the aid of using the wavelet transform, the approximation picture (low–frequency band) and thing image (high-frequency band) may additionally have one-of-a-type bodily meaning. In this project, a new fusion rule is proposed to feature a wavelet coefficient that treats low-frequency and high-frequency bands with high-quality fusion schemes.

Demoiring Images using Efficient Attention Fusion module in Wavelet Domain

When we use a capturing device to capture digital displays, we obtain a vivid rainbow pattern. It's known as "moiré" and it has an impact on image quality and subsequent processing. There are several ways to save money. Modern approaches for eliminating moiré patterns rely on down sampling pooling layers to retrieve multiscale information, which might result in data loss. To show this issue, this research provides a wavelet-based demoiréing approach. Both Inverse Discrete Wavelet Transform (IDWT) and Discrete Wavelet Transform (DWT) are used in this. This technology may efficiently enhance the network related field without any data loss in place of standard down sampling and up sampling. Furthermore, this approach employs an effective attention fusion module to reconstruct additional details of moiré patterns (EAFM). Using an aggregate of green channel attention (ECA), spatial attention (SA), and neighborhood residual getting to know, this module can self-adaptively study several weights of function statistics at distinct degrees and encourage the community to cognizance extra on required statistics as moiré statistics to beautify the getting to know and demoiréing paintings performed via way of means of enhancing photograph excellent and offering the specified Accurate output. Extensive trials the usage of publicly to be had datasets have established that this approach can perpetually cast off moiré styles even as additionally being quantitatively and qualitatively sound.

Fetal phonocardiogram signals denoising using improved complete ensemble (EMD) with adaptive noise and optimal thresholding of wavelet coefficients.

Although fetal phonocardiogram (fPCG) signals have become a good indicator for discovered heart disease, they may be contaminated by various noises that reduce the signals quality and the final diagnosis decision. Moreover, the noise may cause the risk of the data to misunderstand the heart signal and to misinterpret it. The main objective of this paper is to effectively remove noisefrom the fPCG signal to make it clinically feasible. So,we proposed a novel noise reduction method based on Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN), wavelet threshold and Crow Search Algorithm (CSA). This noise reduction method, named ICEEMDAN-DWT-CSA, has three major advantages. They were, (i) A better suppress of mode mixing and a minimized number of IMFs, (ii) A choice of wavelet corresponding to the study signal proven by the literature and (iii) Selection of the optimal threshold value. Firstly, the noisy fPCG signal is decomposed into Intrinsic Mode Functions (IMFs) by the (ICEEMDAN). Each noisy IMFs were decomposed by the Discrete Wavelet Transform (DWT). Then, the optimal threshold value using the (CSA) technique is selected and the thresholding function is carried out in the detail's coefficients. Secondly, each denoised (IMFs) is reconstructed by applying the Inverse Discrete Wavelet Transform (IDWT). Finally, all these denoised (IMFs) are combined to get the denoised fPCG signal. The performance of the proposed method has been evaluated by Signal to Noise Ratio (SNR), Mean Square Error (MSE) and the Correlation Coefficient (COR). The experiment gave a better result than some standard methods.