Enhanced Slime Mould Algorithm for Multilevel Thresholding Image Segmentation Using Entropy Measures

Image segmentation is a fundamental but essential step in image processing because it dramatically influences posterior image analysis. Multilevel thresholding image segmentation is one of the most popular image segmentation techniques, and many researchers have used meta-heuristic optimization algorithms (MAs) to determine the threshold values. However, MAs have some defects; for example, they are prone to stagnate in local optimal and slow convergence speed. This paper proposes an enhanced slime mould algorithm for global optimization and multilevel thresholding image segmentation, namely ESMA. First, the Levy flight method is used to improve the exploration ability of SMA. Second, quasi opposition-based learning is introduced to enhance the exploitation ability and balance the exploration and exploitation. Then, the superiority of the proposed work ESMA is confirmed concerning the 23 benchmark functions. Afterward, the ESMA is applied in multilevel thresholding image segmentation using minimum cross-entropy as the fitness function. We select eight greyscale images as the benchmark images for testing and compare them with the other classical and state-of-the-art algorithms. Meanwhile, the experimental metrics include the average fitness (mean), standard deviation (Std), peak signal to noise ratio (PSNR), structure similarity index (SSIM), feature similarity index (FSIM), and Wilcoxon rank-sum test, which is utilized to evaluate the quality of segmentation. Experimental results demonstrated that ESMA is superior to other algorithms and can provide higher segmentation accuracy.

Review on Image Recoloring Methods for Efficient Naturalness by Coloring Data Modeling Methods for Low Visual Deficiency

Recent research has discovered new applications for object tracking and identification by simulating the colour distribution of a homogeneous region. The colour distribution of an object is resilient when it is subjected to partial occlusion, scaling, and distortion. When rotated in depth, it may remain relatively stable in other applications. The challenging task in image recoloring is the identification of the dichromatic color appearance, which is remaining as a significant requirement in many recoloring imaging sectors. This research study provides three different vision descriptions for image recoloring methods, each with its own unique twist. The descriptions of protanopia, deuteranopia, and tritanopia may be incorporated and evaluated using parametric, machine learning, and reinforcement learning techniques, among others. Through the use of different image recoloring techniques, it has been shown that the supervised learning method outperforms other conventional methods based on performance measures such as naturalness index and feature similarity index (FSIM).

Toward on-the-fly trajectory optimization for C-arm CBCT under strong kinematic constraints

Cone beam computed tomography (CBCT) has become a vital tool in interventional radiology. Usually, a circular source-detector trajectory is used to acquire a three-dimensional (3D) image. Kinematic constraints due to the patient size or additional medical equipment often cause collisions with the imager while performing a full circular rotation. In a previous study, we developed a framework to design collision-free, patient-specific trajectories for the cases in which circular CBCT is not feasible. Our proposed trajectories included enough information to appropriately reconstruct a particular volume of interest (VOI), but the constraints had to be defined before the intervention. As most collisions are unpredictable, performing an on-the-fly trajectory optimization is desirable. In this study, we propose a search strategy that explores a set of trajectories that cover the whole collision-free area and subsequently performs a search locally in the areas with the highest image quality. Selecting the best trajectories is performed using simulations on a prior diagnostic CT volume which serves as a digital phantom for simulations. In our simulations, the Feature SIMilarity Index (FSIM) is used as the objective function to evaluate the imaging quality provided by different trajectories. We investigated the performance of our methods using three different anatomical targets inside the Alderson-Rando phantom. We used FSIM and Universal Quality Image (UQI) to evaluate the final reconstruction results. Our experiments showed that our proposed trajectories could achieve a comparable image quality in the VOI compared to the standard C-arm circular CBCT. We achieved a relative deviation less than 10% for both FSIM and UQI metrics between the reconstructed images from the optimized trajectories and the standard C-arm CBCT for all three targets. The whole trajectory optimization took approximately three to four minutes.

A Generative Adversarial Network-Based Image Denoiser Controlling Heterogeneous Losses

We propose a novel generative adversarial network (GAN)-based image denoising method that utilizes heterogeneous losses. In order to improve the restoration quality of the structural information of the generator, the heterogeneous losses, including the structural loss in addition to the conventional mean squared error (MSE)-based loss, are used to train the generator. To maximize the improvements brought on by the heterogeneous losses, the strength of the structural loss is adaptively adjusted by the discriminator for each input patch. In addition, a depth wise separable convolution-based module that utilizes the dilated convolution and symmetric skip connection is used for the proposed GAN so as to reduce the computational complexity while providing improved denoising quality compared to the convolutional neural network (CNN) denoiser. The experiments showed that the proposed method improved visual information fidelity and feature similarity index values by up to 0.027 and 0.008, respectively, compared to the existing CNN denoiser.

A New Harris Hawks-Cuckoo Search Optimizer for Multilevel Thresholding of Thermogram Images

The exploitation capability of the Harris Hawks optimization (HHO) is limited. This problem is solved here by incorporating features of Cuckoo search (CS). This paper proposes a new algorithm called Harris hawks-cuckoo search (HHO-CS) algorithm. The algorithm is validated using 23 Benchmark functions. A statistical analysis is carried out. Convergence of the proposed algorithm is studied. Nonetheless, converting color breast thermogram images into grayscale for segmentation is not effective. To overcome the problem, we suggest an RGB colour component based multilevel thresholding method for breast cancer thermogram image analysis. Here, 8 different images from the Database for Research Mastology with Infrared images are considered for the experiments. Both 1D Otsu’s between-class variance and Kapur's entropy are considered for a fair comparison. Our proposal is evaluated using the performance metrics – Peak Signal to Noise Ratio (PSNR), Feature Similarity Index (FSIM), Structure Similarity Index (SSIM). The suggested method outperforms the grayscale based multilevel thresholding method proposed earlier. Moreover, our method using 1D Otsu’s fitness functions performs better than Kapur’s entropy based approach. The proposal would be useful for analysis of infrared images. Finally, the proposed HHO-CS algorithm may be useful for function optimization to solve real world engineering problems.

Elephant Herding Optimization for Multi-Level Image Thresholding

Multilevel thresholding plays a significant role in the arena of image segmentation. The main issue of multilevel image thresholding is to select the optimal combination of threshold value at different level. However, this problem has become challenging with the higher number of levels, because computational complexity is increased exponentially as the increase of number of threshold. To address this problem, this paper has proposed elephant herding optimization (EHO) based multilevel image thresholding technique for image segmentation. The EHO method has been inspired by the herding behaviour of elephant group in nature. Two well-known objective functions such as ‘Kapur's entropy' and ‘between-class variance method' have been used to determine the optimized threshold values for segmentation of different objects from an image. The performance of the proposed algorithm has been verified using a set of different test images taken from a well-known benchmark dataset named Berkeley Segmentation Dataset (BSDS). For comparative analysis, the results have been compared with three popular algorithms, e.g. cuckoo search (CS), artificial bee colony (ABC) and particle swarm optimization (PSO). It has been observed that the performance of the proposed EHO based image segmentation technique is efficient and promising with respect to the others in terms of the values of optimized thresholds, objective functions, peak signal-to-noise ratio (PSNR), structure similarity index (SSIM) and feature similarity index (FSIM). The algorithm also shows better convergence profile than the other methods discussed.

Symbiotic Organisms Search Optimization for Multilevel Image Thresholding

Determination of optimum thresholds is the prime concern of any multilevel image thresholding technique. The traditional methods for multilevel thresholding are computationally expensive, time-consuming, and also suffer from lack of accuracy and stability. To address this issue, the authors propose a new methodology for multilevel image thresholding based on a recently developed meta-heuristic algorithm, Symbiotic Organisms Search (SOS). The SOS algorithm has been inspired by the symbiotic relationship among the organism in nature. This article has utilized the concept of the symbiotic relationship among the organisms to optimize three objective functions: Otsu's between class variance and Kapur's and Tsallis entropy for image segmentation. The performance of the SOS based image segmentation algorithm has been evaluated using a set of benchmark images and has been compared with four recent meta-heuristic algorithms. The algorithms are compared in terms of effectiveness and consistency. The quality of the algorithms has been estimated by some well-defined quality metrics such as peak signal-to-noise ratio (PSNR), structure similarity index (SSIM), and, feature similarity index (FSIM). The experimental results of the algorithms reveal that the balance of intensification and diversification of the SOS algorithm to achieve the global optima is better than others.

Fuzzy Based Combinatorial Filter Model for Drusen Enhancement in Retinal Fundus Images

This article presents an approach to enhance drusen regions in retinal fundus image of a patient having Age-related Macular Degeneration (AMD). In this approach, a new filter model is developed by combining two processes on images which is named as Combinatorial Filter Model (CFM). The first process is based on processing drusen significant image bit planes and the second process is based on implementing fuzzy inference system (FIS) on bit planes. When bit planes are independently processed, the result improves the visibility of drusen features. An FIS is constructed to process the bit planes to further enhance the processed image. This approach is tested on images with drusen features on good quality images in proprietary database and comparatively low quality images from STARE database. The objective study of this model shows that drusen elements are enhanced and validated to a 95% significant level. The quality of the enhanced image is evaluated for preservation of drusen features using a proven feature similarity index technique with 0.99 quality index values

Weighted lp Norm Sparse Error Constraint Based ADMM for Image Denoising

In the process of image denoising, the accurate prior knowledge cannot be learned due to the influence of noise. Therefore, it is difficult to obtain better sparse coefficients. Based on this consideration, a weighted lp norm sparse error constraint (WPNSEC) model is proposed. Firstly, the suitable setting of power p in the lp norm is made a detailed analysis. Secondly, the proposed model is extended to color image denoising. Since the noise of RGB channels has different intensities, a weight matrix is introduced to measure the noise levels of different channels, and a multichannel weighted lp norm sparse error constraint algorithm is proposed. Thirdly, in order to ensure that the proposed algorithm is tractable, the multichannel WPNSEC model is converted into an equality constraint problem solved via alternating direction method of multipliers (ADMM) algorithm. Experimental results on gray image and color image datasets show that the proposed algorithms not only have higher peak signal-to-noise ratio (PSNR) and feature similarity index (FSIM) but also produce better visual quality than competing image denoising algorithms.

An Adaptive Curvelet Based Semi-Fragile Watermarking Scheme for Effective and Intelligent Tampering Classification and Recovery of Digital Images

A novel adaptive semi-fragile watermarking scheme for tamper detection and recovery of digital images is proposed in this paper. This scheme involves embedding of content and chroma watermarks generated from the first level Discrete Curvelet Transform (DCLT) coarse coefficients. Embedding is performed by quantizing the first level coarse DCLT coefficients of the input image and amount of quantization is intelligently decided based on the energy contribution of the coefficients. During watermark extraction, a tampered matrix is generated by comparing the feature similarity index value between each block of extracted and generated watermarks. The tampered objects are subsequently identified and an intelligent report is formed based on their severity classes. The recovery of the tampered objects is performed using the generated DCLT coefficients from luminance and chrominance components of the watermarked image. Results reveal that the proposed method outperforms existing method in terms of tamper detection and recovery of digital images.