Image colorization: A survey and dataset

At present, deep learning method is widely used in the field of gray image colorization. Qinghai farmer painting has distinct national characteristics. The farmer painting has bright colors, high saturation, chaotic color distribution and low color contrast, so it is difficult to restore the image color with high fidelity by using the general deep learning image colorization method. The Pix2Pix generation adversarial network of grayscale image colorization method uses the Leaky ReLU function as the activation function. The proposal algorithm replaces the maximum pooling layer with the convolution layer to retain more image feature information and further to improve the color simulation effect. Meanwhile, in view of the lack of relevant Qinghai farmer painting data set, the data set of Qinghai farmer paintings is constructed to meet the needs of network training. The experimental results show that the improved method further improves the color effect and can generate high quality color images of Qinghai farmer paintings with more real color distribution.

Diving deeper into underwater image enhancement: A survey

In order to overcome the lack of the multispectral image (MS) and adequately preserve the spatial information of panchromatic (PAN) image and the spectral information of MS image, this study proposes a method which adds the spectral information of the prior MS to the prior PAN during training, and only the posterior PAN is needed for predicting. Firstly, we introduce the autoencoder model based on image colorization and discuss its feasibility in the field of multi-band remote sensing image pan-sharpening. Then, the image quality evaluation functions including spatial and spectral indexes are formed as the loss function to control the image colorization model. Because the loss function contains spatial and spectral evaluation indexes, it could directly calculate the loss between the network output and the label considering characteristics of remote sensing images. Besides, the training data in our model is original PAN, this means that it is not necessary to make the simulated degraded MS and PAN data for training which is a big difference from most existing deep learning pan-sharpening methods. The new loss function including the spectral and spatial quality instead of the general MSE (mean square error), only the original PAN instead of the simulated degraded MS + PAN to be inputted, only the spectral feature instead of the direct fusion result to be learned, these three aspects change the current learning framework and optimization rule of deep learning pan-sharpening. Finally, thousands of remote sensing images from different scenes are adopted to make the training dataset to verify the effectiveness of the proposed method. In addition, we selected seven representative pan-sharpening algorithms and four widely recognized objective fusion metrics to evaluate and compare the performance on the WorldView-2 experimental data. The results show that the proposed method achieves optimal performance in terms of both the subjective visual effect and the object assessment.

Colorization is a process of adding colors to a black and white image. There exists vast amounts of illustrations and photographs that lack color information such as modern medical imaging to antique photography. Visual appeal and expressiveness of these black and white images can be improved by adding colors. Image colorization is a challenging topic of current researches in Computer Visions. Significant user interaction is required in the form of placing numerous color scribbles, looking at related images etc for coloring the image in traditional or manual coloring of image. This is difficult because of the complex shades of colors present in the image. Automating the image colorization methods has various benefits in different field such as medical, art, entertainment etc. Applying machine learning techniques to this process reduces the human efforts substantially. Image colorization is the process of taking an input grayscale image and then producing an output colorized image. In this proposed chapter a technique of colorization of grayscale image using deep learning is discussed.

We propose an image colorization method using fast soft-thresholding of singular values (singular value thresholding). An image colorization method with nuclear norm minimization (NNM) has been proposed and brings good results. NNM usually requires iterative application of singular value decomposition (SVD) for singular value thresholding. However, the computational cost of SVD in the colorization method becomes too expensive to handle high-resolution images. In this paper, we reduce its computational cost by using Chebyshev polynomial approximation (CPA). Singular value thresholding is expressed by a multiplication of certain matrices derived from the characteristic of CPA. As a result, our CPA-based technique makes the image colorization method much more efficient. In addition, we replace the optimization method used in the image colorization method by alternating direction method of multipliers, which further accelerates the computation. Experimental results verify the effectiveness of our method with respect to the computation time and the approximation precision.

This article proposes a novel approach for predicting depth from a single image in a self-supervised manner, specifically using generative adversarial networks (GANs) with enhancements to improve training performance. We train our model to generate disparity maps or inverse depth maps using left/right stereo image supervision rather than direct supervision from depth ground truth. We avoid the problem of training instability in standard GANs by employing a relativistic discriminator that evaluates how realistic an input image is. Since the reconstructed image quality determines the disparity map's accuracy, we adopt the GMSD, one of the standout IQA algorithms, as a part of the image reconstruction loss function. A VGG-19 pre-trained model-based perceptual loss is used to optimize our model in a feature space instead of pixel space and to enhance reconstructed images' perceptual quality with the learning-based perspective. Also, we add another symmetric generator that takes the right images as input, which helps to maintain the consistency of left/right views and enrich our training data with reasonable overhead on the training time. Our network is trained to predict depths of 640 × 192 images using the KITTI stereo dataset, which shows an outstanding performance compared to state-of-the-art works in various evaluation metrics. We verify our method's modules by performing an ablation study and evaluate the generalization performance on CityScapes and Make3D datasets.

Automatic image colorization has made tremendous progress in recent years. However, previous methods rarely consider the inherent diversity of natural images, and existing diverse colorization approaches still suffer from the following two limitations. i) The diversity of generated images is limited. ii) Some inevitable artifacts substantially degrade the quality of colorization results. To cope with such problems, we propose a novel diverse image colorization network based on vanilla GAN, which extracts the deep color priors through an initial colorization network first, and then utilizes these priors to modulate the diverse generation to achieve high-fidelity colorization outputs. In addition, a novel triplet latent regularization is proposed to constrain the correspondence between latent codes and generated images, which efficiently alleviates mode collapse and encourages more diverse colorization results. Extensive experiments on three benchmark datasets demonstrate the superiority of our method over existing diverse image colorization models.

A survey and analysis on automatic image annotation

Image captioning has attracted more and more attention in the field of artificial intelligence. It is a challenging task that involves natural language description generation by automatic visual data analysis. This involves understanding the visual contents of the image and then generating natural language sentences describing the image. A number of techniques have been proposed to solve this problem. This chapter presents a survey on deep learning-based image captioning techniques and the recent advances in this field. An exhaustive explanation of the evaluation metrics is also presented. The focus of this chapter is attention-guided methods since they have shown promising results for image caption generation. The existing evaluation methods are discussed in detail, along with examples to elaborate the mechanism behind them. Furthermore, the most used benchmark datasets – Flickr8k, Flick30k, and MS-COCO – are also described. Image captioning has applications in many areas. In this chapter, the application of image captioning on industrial images is presented.

Since low-light-level images are generally grayscale images that lack color information, it is necessary to colorize these images to enhance human vision for night vision scene. In this paper, a low-light-level image colorization method is proposed based on Laws’ texture feature descriptor. Laws’ filter can analyze the texture features of the image, which are the pointers we used to find the corresponding color information for low-light-level gray images from the reference color image. The experimental results demonstrate that our colorization method can make night vision images have colors closer to natural perception and help observers better understand night vision scenes.

Multi-dimensional classification with Bayesian networks

Colorization is a process of converting grayscale images into visually acceptable colorized images. It is a complex problem because no unique solution exists, i.e., the objects within the image cannot be associated with one color only. Colorized image quality assessment is also a complex problem because in real life the reference image for comparison with colorized results does not exist. The main purpose of colorization is convincing the observer in the credibility of the colorized image, disregarding color accuracy. For that reason, both subjective and objective evaluation of colorization results show imperfections. In this paper, the objective quality evaluation of colorized images has been conducted with five existing metrics and one new metric. The database of colorized images has been obtained by deep learning colorization methods. Based on the conducted subjective testing, a correlation between objective metrics and subjective grades has been determined. The results indicate that the improvement of objective metrics for the evaluation of colorized images is needed.

The Internet of Things (IoT) has transformed many aspects of modern life, from healthcare and transportation to home automation and industrial control systems. However, the increasing number of connected devices has also led to an increase in security threats, particularly from botnets. To mitigate these threats, various machine learning (ML) and deep learning (DL) techniques have been proposed for IoT botnet attack detection. This systematic review aims to identify the most effective ML and DL techniques for detecting IoT botnets by delving into benchmark datasets, evaluation metrics, and data pre-processing techniques in detail. A comprehensive search was conducted in multiple databases for primary studies published between 2018 and 2023. Studies were included if they reported the use of ML or DL techniques for IoT botnet detection. After screening 1,567 records, 25 studies were included in the final review. The findings suggest that ML and DL techniques show promising results in detecting IoT botnet attacks, outperforming traditional signature-based methods. However, the effectiveness of the techniques varied depending on the dataset, features, and evaluation metrics used. Based on the synthesis of the findings, this review proposes a taxonomy for ML and DL techniques in IoT botnet attack detection and provides recommendations for future research in this area. This review illuminates the considerable potential of ML and DL approaches in bolstering the detection of IoT botnet attacks, thereby offering valuable insights to researchers involved in the domain of IoT security.

An improved DualGAN for near-infrared image colorization

A survey on deep learning-based image forgery detection