Image Pixels Research Articles

Statistical biases correction in channelized Hotelling model observers

Channelized Hotelling model observers are efficient at simulating the human observer visual performance in medical imaging detection tasks. However, channelized Hotelling observers (CHO) are subject to statistical biases from zero-signal and finite-sample effects. The point estimate of the d' value is also not always symmetric with exact confidence interval (CI) bounds determined for the infinitely trained CHO. A method for correcting these statistical biases and CI asymmetry is studied. CHO d' values and CI bounds with hold-out and resubstitution methods were computed for a range of 200x200 pixels images from 20 to 10 000 images for 10, 40 and 96 channels from a set of 20 000 images with gaussian coloured simulated noise and simulated signal. The median of the non-central F cumulative distribution (F'), which is the CHO underlying statistical behaviour for the resubstitution method, was computed, and compared to d' values and CI bounds. A set of experimental data was used to evaluate F' median values. The F' median allows to get accurate corrected simulated d' values down to zero-signals. For small d' values, the variation of d' values with the inverse of number of images is not linear while the F' median allows a good correction in such conditions. The F' median is also inherently symmetric with regards to the confidence interval. With experimental data, F' median values in a range of about 1 to 10 d' values were within -0.8% to 4.7% of linearly extrapolated values at an infinite number of images. The F' median correction is an effective simultaneous correction of the zero-signal statistical bias and finite-sample statistical bias, and of confidence interval asymmetry of channelized Hotelling observers.&#xD.

Optimisation of semantic segmentation algorithm for autonomous driving using U-NET architecture

In autonomous driving systems, the semantic segmentation task involves scene partition into numerous expressive portions by classifying and labelling every image pixel for semantics. The algorithm used for semantic segmentation has a vital role in autonomous driving architecture. This paper's main contribution is optimising the semantic segmentation algorithm for autonomous driving by modifying the U-NET architecture. The optimisation techniques involve five different methods, which include; no batch normalisation network, with batch normalisation network, network with reduction in filters, average ensemble network, and weighted average ensemble network. The validation accuracy observed for the five methods were 90.28%, 91.68%, 89.80%, 92.04%, and 92.21% respectively. By reducing the filters in the network, the computation time reduces (Epoch time: 1 s 64 ms/step) as opposed to the typical (Epoch time: 4 s 260 ms/step), but the accuracy reduces. The optimisation techniques were evaluated for metrics like mean intersection over union (IoU), IoU for class, dice-metric, dice_coefficient_loss, validation loss, and accuracy. The dataset of 300 images used for this paper's study was generated using the open-source car learning to act (CARLA) simulator platform.

Depth-Wise Convolutions in Vision Transformers for efficient training on small datasets

The Vision Transformer (ViT) leverages the Transformer’s encoder to capture global information by dividing images into patches and achieves superior performance across various computer vision tasks. However, the self-attention mechanism of ViT captures the global context from the outset, overlooking the inherent relationships between neighboring pixels in images or videos. Transformers mainly focus on global information while ignoring the fine-grained local details. Consequently, ViT lacks inductive bias during image or video dataset training. In contrast, convolutional neural networks (CNNs), with their reliance on local filters, possess an inherent inductive bias, making them more efficient and quicker to converge than ViT with less data. In this paper, we present a lightweight Depth-Wise Convolution module as a shortcut in ViT models, bypassing entire Transformer blocks to ensure the models capture both local and global information with minimal overhead. Additionally, we introduce two architecture variants, allowing the Depth-Wise Convolution modules to be applied to multiple Transformer blocks for parameter savings, and incorporating independent parallel Depth-Wise Convolution modules with different kernels to enhance the acquisition of local information. The proposed approach significantly boosts the performance of ViT models on image classification, object detection, and instance segmentation by a large margin, especially on small datasets, as evaluated on CIFAR-10, CIFAR-100, Tiny-ImageNet and ImageNet for image classification, and COCO for object detection and instance segmentation. The source code can be accessed at https://github.com/ZTX-100/Efficient_ViT_with_DW.

Egg mass classification considering the hatching process of Pomacea canaliculata

Pomacea canaliculata feeds on seedlings that have been planted less than three weeks ago. This study aimed to construct an imaging system that can eliminate the egg masses of P. canaliculata before they hatch and multiply. An image classification method is proposed that can recognize the state of hatching of the egg masses. As hatching process proceeds, this state changes from “freshly laid” to “maturing” and then to “mature.” In the proposed method, first, egg image pixels are detected using a four-label semantic segmentation model that includes the background label. Next, the egg masses are classified by analyzing the distribution of the labeled pixels in the egg masses. We conducted an experiment in which we verified the effectiveness of the proposed method on images of egg masses from an agricultural canal and evaluated its classification accuracy. The F1-score of the proposed method was 1.00 when the weather was cloudy and 0.842 when it was sunny, demonstrating that the state of hatching could be identified accurately regardless of the brightness of the day. Using this classification method, only newly laid eggs can be dropped into water and eliminated, which is a step forward in the automation of this method for P. canaliculata control.

An seamless stitching method for large field equivalent center projection image based on rotating camera

Digital cameras are limited by a narrow field of view and a large photosensitive unit, resulting in images with a small frame size and low resolution. This reduces the acquisition range and measurement accuracy of stereo vision in close-range photogrammetry, making it difficult to meet the requirements for precise close-range photogrammetry in high-precision industrial engineering fields, and limiting the significant development of digital close-range photogrammetry. For this reason, based on the characteristics of ground close-range photogrammetry, this paper proposes a large-format image acquisition method for rotating cameras. By designing a simple and structurally relaxed rotating camera, a rigorous seamless stitching model for large-format images is constructed, forming a large-format equivalent central projection image acquisition mechanism that meets the requirements of precise close-range photogrammetry. Finally, the effectiveness of the proposed method is verified through experiments. The results show that the proposed method effectively increases the coverage of a single camera station. The large-format image obtained through three degrees of rotation increases the image size from 916 × 687 pixels in a single image to 4977 × 671 pixels in a large-format image. This method solves the problem of the small view field of digital cameras, complementing the theory of precision close-range photogrammetry and providing necessary theoretical support for technological development in the field of precision industrial engineering.

Introduction to matrix-based method for analyzing hybrid multidimensional prostate MRI data.

A new approach to analysis of prostate hybrid multidimensional MRI (HM-MRI) data was introduced in this study. HM-MRI data were acquired for a combination of a few echo times (TEs) and a few b-values. Naturally, there is a matrix associated with HM-MRI data for each image pixel. To process the data, we first linearized HM-MRI data by taking the natural logarithm of the imaging signal intensity. Subsequently, a hybrid symmetric matrix was constructed by multiplying the matrix for each pixel by its own transpose. The eigenvalues for each pixel could then be calculated from the hybrid symmetric matrix. In order to compare eigenvalues between patients, three b-values and three TEs were used, because this was smallest number of b-values and TEs among all patients. The results of eigenvalues were displayed as qualitative color maps for easier visualization. For quantitative analysis, the ratio (λr) of eigenvalues (λ1, λ2, λ3) was defined as λr=(λ1/λ2)/λ3 to compare region of interest (ROI) between prostate cancer (PCa) and normal tissue. The results show that the combined eigenvalue maps show PCas clearly and these maps are quite different from apparent diffusion coefficient (ADC) and T2 maps of the same prostate. The PCa has significant larger λr, smaller ADC and smaller T2 values than normal prostate tissue (p<0.001). This suggests that the matrix-based method for analyzing HM-MRI data provides new information that may be clinically useful. The method is easy to use and could be easily implemented in clinical practice. The eigenvalues are associated with combination of ADC and T2 values, and could aid in the identification and staging of PCa.

APPLICATION OF THE ITEM RESPONSE THEORY IN IMAGE PROCESSING

The Item Response Theory (IRT) was applied to image processing where the item's difficulty parameter was used to reconstruct a region of interest in an image contaminated with noise. Shannon's entropy was used for the dichotomization of the image pixels. To demonstrate the efficiency of the proposed method, it was used on simulated data that can be related to functional magnetic resonance imaging (fMRI). The results on simulated data showed that IRT achieved a high percentage of accuracy in identifying the region of interest in the image and that, on average, the estimation converges to the true region of interest, thus proving to be a promising method for the analysis of such data.

Research on cloud dynamic public key information security based on elliptic curve and primitive Pythagoras

In order to solve the problems of key redundancy, transmission and storage security and the difficulty of realizing the one-time-secret mode in image encryption algorithm, this paper proposes a cloud dynamic public key information security scheme based on elliptic curve and primitive Pythagoras. The 6-D Lorentz hyperchaos system is used as the key generator, and the generated key is stored in the cloud key management center, which supports automatic update and dynamic change to further ensure the security of the key. The encryption key is selected by elliptic curve and primitive Pythagorean triplet. In the scrambling algorithm, according to the parity of random number and the parity of image pixel value coordinates, elliptic curve is used to reset the plaintext image. In the diffusion algorithm, the three-side rotation characteristic of the Rubik’s cube is simulated, the binary sequence of pixel values is stored at eight vertices of the cube, the cube is rotated according to the key value, and the changed 8 new binary numbers are XOR operated with the random sequence to obtain the new pixel value. The algorithm in this paper is applied to network transmission, which can effectively prevent data eavesdropping, tampering and forgery, ensure communication security.

Discrete Cosine Transform-Based Joint Spectral–Spatial Information Compression and Band-Correlation Calculation for Hyperspectral Feature Extraction

Prediction tasks over pixels in hyperspectral images (HSI) require careful effort to engineer the features used for learning a classifier. However, the generated classification map may suffer from an over-smoothing problem, which is manifested in significant differences from the original image in terms of object boundaries and details. To address this over-smoothing problem, we designed a method for extracting spectral–spatial-band-correlation (SSBC) features. In SSBC features, joint spectral–spatial feature extraction is considered a discrete cosine transform-based information compression, where a flattening operation is used to avoid the high computational cost induced by the requirement of distillation from 3D images for joint spectral–spatial information. However, this process can yield extracted features with lost spectral information. We argue that increasing the spectral information in the extracted features is the key to addressing the over-smoothing problem in the classification map. Consequently, the normalized difference vegetation index and iron oxide are improved for HSI data in extracting band-correlation features as added spectral information because their calculations, involving two spectral bands, are not appropriate for the abundant spectral bands of HSI. Experimental results on four real HSI datasets show that the proposed features can significantly mitigate the over-smoothing problem, and the classification performance is comparable to that of state-of-the-art deep features.

The implementation of algebraic complex lookup tables over non-chain Galois ring extensions and Henon map in multimedia security

Abstract Image encryption is crucial for web-based data storage and transmission. Complex algebraic structures play a vital role in providing unique features and binary operations. However, current algebraic-based techniques face challenges due to limited key space. To tackle this issue, our study uniquely connects the algebraic structures with a chaotic map. The study introduces a complex non-chain Galois ring structure and a 12-bit substitution box for image substitution. An affine map is utilized to permute image pixels, and the 12-bit substitution box is uniquely mapped to a Galois field for encryption. A two-dimensional Henon map is employed to generate different keys for the XOR operation, resulting in an encrypted image. The resilience of the scheme against various attacks is evaluated using statistical, differential, and quality measures, showcasing its effectiveness against well-known attacks.

MRA-VSS: A Matrix-Based Reversible and Authenticable Visual Secret-Sharing Scheme Using Dual Meaningful Images

Reversible data hiding (RDH) is an approach that emphasizes the imperceptibility of hidden confidential data and the restoration of the original cover image. To achieve these objectives at the same time, in this paper, we design a matrix-based crossover data hiding strategy and then propose a novel matrix-based RDH scheme with dual meaningful image shadows, called MRA-VSS (matrix-based reversible and authenticable visual secret-sharing). Each pixel in a secret image is divided into two parts, and each part is embedded into a cover pixel pair by referring to the intersection point of four overlapping frames. During the share construction phase, not only partial information of the pixel in a secret image but also authentication codes are embedded into the corresponding cover pixel pair. Finally, two meaningful image shadows are derived. The experimental results confirm that our designed MRA-VSS successfully embeds pixels’ partial information and authentication code into cover pixel pairs at the cost of slight distortion during data hiding. Nevertheless, the robustness of our scheme under the steganalysis attack and the authentication capability of our scheme are also proven.

Detection of the Pigment Distribution of Stacked Matcha During Processing Based on Hyperspectral Imaging Technology

Color is a key indicator for evaluating the quality of tea during processing; various processing procedures can significantly affect the content of fat-soluble pigments of tea, which in turn affects the color and quality of finished tea. Therefore, there is an urgent demand for the fast, non-destructive detection of pigments of stacked tea during processing. This paper presents the use of hyperspectral imaging technology (HSI), combined with machine learning algorithms, to detect chlorophyll a, chlorophyll b, and carotenoids in stacked matcha tea during processing. Firstly, a quantitative relationship between HSI data of tea and their pigment contents was developed based on regression analysis, and the results showed that exceptional prediction performance was achieved by the partial least squares regression (PLSR) algorithm combined with the feature band algorithm of competitive adaptive reweighting (CARS), and the Rp2 values of detection models of chlorophyll a, chlorophyll b and carotenoids were 0.90465, 0.92068 and 0.62666, respectively. Then, these quantitative detection models were extended to each pixel in hyperspectral images, achieving point-by-point prediction of pigment components, so the distribution of pigments of stacked tea leaves during processing procedures was successfully visualized on the processing line in situ. By integrating a hyperspectral imaging system into the real-world environment, operators can monitor pigment levels in real time and thus dynamically adjust processing parameters based on real-time data. This study enhances pigment detection efficiency in tea processing, supports process optimization, and aids in quality control.

Asymmetric Large Kernel Distillation Network for efficient single image super-resolution.

Recently, significant advancements have been made in the field of efficient single-image super-resolution, primarily driven by the innovative concept of information distillation. This method adeptly leverages multi-level features to facilitate high-resolution image reconstruction, allowing for enhanced detail and clarity. However, many existing approaches predominantly emphasize the enhancement of distilled features, often overlooking the critical aspect of improving the feature extraction capabilities of the distillation module itself. In this paper, we address this limitation by introducing an asymmetric large-kernel convolution design. By increasing the size of the convolution kernel, we expand the receptive field, which enables the model to more effectively capture long-range dependencies among image pixels. This enhancement significantly improves the model's perceptual ability, leading to more accurate reconstructions. To maintain a manageable level of model complexity, we adopt a lightweight architecture that employs asymmetric convolution techniques. Building on this foundation, we propose the Lightweight Asymmetric Large Kernel Distillation Network (ALKDNet). Comprehensive experiments conducted on five widely recognized benchmark datasets-Set5, Set14, BSD100, Urban100, and Manga109-indicate that ALKDNet not only preserves efficiency but also demonstrates performance enhancements relative to existing super-resolution methods. The average PSNR and SSIM values show improvements of 0.10 dB and 0.0013, respectively, thereby achieving state-of-the art performance.

Enhanced YOLOv8-Based Model with Context Enrichment Module for Crowd Counting in Complex Drone Imagery

Crowd counting in aerial images presents unique challenges due to varying altitudes, angles, and cluttered backgrounds. Additionally, the small size of targets, often occupying only a few pixels in high-resolution images, further complicates the problem. Current crowd counting models struggle in these complex scenarios, leading to inaccurate counts, which are crucial for crowd management. Moreover, these regression-based models only provide the total count without indicating the location or distribution of people within the environment, limiting their practical utility. While YOLOv8 has achieved significant success in detecting small targets within aerial imagery, it faces challenges when directly applied to crowd counting tasks in such contexts. To overcome these challenges, we propose an improved framework based on YOLOv8, incorporating a context enrichment module (CEM) to capture multiscale contextual information. This enhancement improves the model’s ability to detect and localize tiny targets in complex aerial images. We assess the effectiveness of the proposed framework on the challenging VisDrone-CC2021 dataset, and our experimental results demonstrate the effectiveness of this approach.

Multi-dimensional dense attention network for pixel-wise segmentation of optic disc in colour fundus images.

Segmentation of retinal fragments like blood vessels, Optic Disc (OD), and Optic Cup (OC) enables the early detection of different retinal pathologies like Diabetic Retinopathy (DR), Glaucoma, etc. Accurate segmentation of OD remains challenging due to blurred boundaries, vessel occlusion, and other distractions and limitations. These days, deep learning is rapidly progressing in the segmentation of image pixels, and a number of network models have been proposed for end-to-end image segmentation. However, there are still certain limitations, such as limited ability to represent context, inadequate feature processing, limited receptive field, etc., which lead to the loss of local details and blurred boundaries. A multi-dimensional dense attention network, or MDDA-Net, is proposed for pixel-wise segmentation of OD in retinal images in order to address the aforementioned issues and produce more thorough and accurate segmentation results. In order to acquire powerful contexts when faced with limited context representation capabilities, a dense attention block is recommended. A triple-attention (TA) block is introduced in order to better extract the relationship between pixels and obtain more comprehensive information, with the goal of addressing the insufficient feature processing. In the meantime, a multi-scale context fusion (MCF) is suggested for acquiring the multi-scale contexts through context improvement. Specifically, we provide a thorough assessment of the suggested approach on three difficult datasets. In the MESSIDOR and ORIGA data sets, the suggested MDDA-NET approach obtains accuracy levels of 99.28% and 98.95%, respectively. The experimental results show that the MDDA-Net can obtain better performance than state-of-the-art deep learning models under the same environmental conditions.

Application of Artificial Intelligence and Image Processing for the Cultivation of Chlorella sp. Using Tubular Photobioreactors.

By integrating innovative technologies to enhance the efficiency and sustainability of production, this study specifies the establishment of a cutting-edge growing system for Chlorella sp. microalgae. Improvement of a system for the real-time, noninvasive observation and management of algae growth employing a closed tubular photobioreactor (PBR) engineered with computational fluid dynamics (CFD), combined with the Internet of things (IoT), artificial intelligence (AI), and image processing technologies was the major goal of this research. The fitting of seven types of sensors to identify key characteristics such as temperature, pH, light intensity, electrical conductivity (EC), flow rate, oxygen content, and light exposure duration was included in the research method. To manage the gaining of sensor data and system operations, an ESP8266 microcontroller was used as the main control unit, while 33 × 33 pixel images were taken with an ESP32 camera at 30 min intervals to assess growth by evaluating color intensity, enabling real-time evaluation of algal density without sampling or disturbing growth. Forecasting and enhancing farming situations was the goal of producing these machine learning (ML) models. Uniformly dispersed between 12 and 24 h light cycles, the data set comprised 602 samples. Considerable improvements were observed in the results for biomass productivity, with constant 24 h lighting yielding a 7.19% increase, counter to a 2.09% increase seen in the 12 h cycle. Temperature and light intensity are the most significant parameters for growth, as revealed by analysis of Feature Importance. The eXtreme Gradient Boosting (XGBoost) model showed remarkable effectiveness in terms of projecting growth, attaining an R 2 value of 0.9997 for the training data set. With important benefits for the development of renewable energy, food supply, and environmental modification in the future, this research highlights the competence of intelligent technology to strengthen microalgae production.

Embedding Quantum Random Phase Encoding Arnold Transform for Advanced Image Security

Purpose: This research proposes an improvised version of the image encryption technique by incorporating Quantum Random Phase Encoding with the Arnold Transform to help enhance the strength and non-predictability of the encryption process. In this research work, some ideas gained from quantum-based methods have been brought to use with conventional approaches in image encryption techniques for enhancing their security. Methods: This model represents the basic methodology that underlies the Arnold Transform for scrambling the arrangement of image pixels to mask recognizable structures within quantum random phase encoding to introduce complexity through quantum-generated random phases. Result: The experimental results show much improvement in encryption efficiency. For example, in the case of "Cameraman" and "Lena", MSE parameters are 98.134 and 104.76, respectively; these now go up to 832.01 and 888.78. This implies that the higher decrement of these values 21.17 dB and 23.98 dB to 13.41 dB and 13.33 dB translates into higher distortion with higher security. Meanwhile, UACI and NPCR are also very steady and the mean value is about 0.3356 to 0.3358 and 99.60 to 99.61, which proves that this method has been effective in changing the pixel's value, and sensitive input changes. Novelty: This work is novel due to the introduction of quantum technologies in the classical methodology of image encryption. While classical techniques make use of conventional transforms for scrambling, like the Arnold Transform, this work embeds quantum randomness and intricacy in the process as a means of encoding namely, Quantum Random Phase Encoding.

Reversible data hiding in encrypted images using Pixel Shifting Approach (PSA)

This study proposes a novel technique for reversibly embedding data within encrypted images, specifically utilizing a Pixel Shifting Approach (PSA). Traditional data hiding methods commonly employ the modification of the least significant bits (LSBs) of pixels in the original image. In contrast, methods using encrypted cover images offer a wider range of data hiding techniques that are not constrained by LSB modifications. The proposed data hiding method embeds data into an encrypted cover image by dividing it into non-overlapping blocks and then permuting pixels within each block according to a defined rule, rather than modifying the LSBs of pixels. The entropy of the image remains preserved after data embedding, making it difficult to detect the location of hidden data. To extract the hidden data, the encrypted image is decrypted, and the smoothness of each block is measured. The block with the lowest smoothness is identified as the one containing the hidden data, which is also the original block. This is because the pixels within each block have different statistical properties compared to other blocks, allowing for data recovery. The proposed method offers the advantage of embedding 4 bits of data per block, surpassing the performance of the existing Modified Pixel Shifting Approach (MPSA). Moreover, the entropy preservation of the stego-image is comparable to MPSA, while the algorithm’s execution time is more efficient. The simpler data hiding rule compared to MPSA enables faster encoding and decoding. Reversibility is guaranteed by preserving the correlation among pixels within each block when extracting data from the stego-image and restoring the original image. Experimental results demonstrate that the proposed method significantly outperforms existing techniques in terms of data hiding efficiency and security.

TRUST-DART: Land Surfaces Temperature and Emissivity Nonlinear Mapping from Nonisothermal Mixed Pixels of Satellite Images with the DART 3D Radiative Transfer Model

Abstract. Coarse spatial resolution of Thermal Infrared (TIR) satellites hampers measuring the temperature and emissivity of scene elements from space that improves our understanding of land surfaces' thermal behavior. The stringent conditions of current TIR unmixing methods hinder the production of extensive component temperature and emissivity products. To address this, we designed a gradient-based multi-pixel physical model, TRUST-DART, to derive the temperature and emissivity of urban features from non-isothermal mixed pixels of satellite images using the DART 3D radiative transfer model. Unlike traditional TIR unmixing methods, TRUSTDART is not constrained by issues related to spatial, spectral, temporal resolution, angular, scene, field measurement requirements, or manual operations. Its inputs include an at-surface radiance image, downwelling sky irradiance, a 3D urban mock-up with feature information, and DART input parameters such as spatial resolution. It generates maps of emissivity and temperature per urban feature. Its accuracy is validated for two vegetation and urban scenes and two types of images (DART simulated pseudo satellite and ASTER observed images). The accuracy of the TRUST-DART depends heavily on the fraction of components. TRUST-DART proves robust for high-fraction components. However, its accuracy decreases with decreasing fractions. TRUST-DART is distributed with DART and is available for education and research via Toulouse III University (https://dart.omp.eu).

Point cloud segmentation method based on an image mask and its application verification

Abstract Accurately perceiving three-dimensional (3D) environments or objects is crucial for the advancement of artificial intelligence interaction technologies. Currently, various types of sensors are employed to obtain point cloud data for 3D object detection or segmentation tasks. While this multi-sensor approach provides more precise 3D data than monocular or stereo cameras, it is also more expensive. The advent of RGB-D cameras, which provide both RGB images and depth information, addresses this issue. In this study, we propose a point cloud segmentation method based on image masks. By using an RGB-D camera to capture color and depth images, we generate image masks through object recognition and segmentation. Given the mapping relationship between RGB image pixels and point clouds, these image masks can be further used to extract the point cloud data of the target objects. The experimental results revealed that the average accuracy of target segmentation was 84.78%, which was close to that of PointNet++. Compared with three traditional segmentation algorithms, the accuracy was improved by nearly 23.97%. The running time of our algorithm is reduced by 95.76% compared to the PointNet++ algorithm, which has the longest running time; and by 15.65% compared to the LCCP algorithm, which has the shortest running time among traditional methods. Compared with PointNet++, the segmentation accuracy was improved. This method addressed the issues of low robustness and excessive reliance on manual feature extraction in traditional point cloud segmentation methods, providing valuable support and reference for the accurate segmentation of 3D point clouds.