China
Africa
Explore chapters and articles related to this topic
Image Super-Resolution: Historical Overview and Future Challenges
Published in Peyman Milanfar, Super-Resolution Imaging, 2017
In most digital imaging applications, high-resolution images or videos are usually desired for later image processing and analysis. The desire for high-resolution stemsfrom two principal application areas: improvement of pictorial information for human interpretation; and helping representation for automatic machine perception. Image resolution describes the details contained in an image, the higher the resolution, the more image details. The resolution of a digital image can be classified in many different ways: pixel resolution, spatial resolution, spectral resolution, temporal resolution, and radiometric resolution. In this context, we are mainly interested in spatial resolution.
Super-Resolution Reconstruction of Infrared Images Adopting Counter Propagation Neural Networks
Published in Chiranji Lal Chowdhary, Intelligent Systems, 2019
Image resolution is of importance to image processing. Higher image resolution holds more details, which is important to image recognition or image segmentation. Thus, it is desired to obtain a high-resolution (HR) image from its low-resolution (LR) counterpart(s). Due to the restrictions of the IR capture device, IR image resolution is generally low when compared with the visible image resolution. Super resolution (SR) can generate a HR image from a LR image or a sequence of LR images. SR methods can be categorized into two classes, that is, multiframe-based SR and learning-based SR. In the learning-based approach, the HR image can be derived from its corresponding LR image with an image database.
Evaluation of Image Quality
Published in Junichi Nakamura, Image Sensors and Signal Processing for Digital Still Cameras, 2017
In practical terms, image resolution is the measure of the size of details that can be depicted by a given image transmission system. It should be clearly distinguished from the sharpness of the whole image or the total evaluation of image quality. Resolution is just a parameter determining whether the output signal can contain enough information to read detail of a given size. It does not have much relation to whether the picture is sharp or whether it is free from interference (for example, color moiré).
Deep remote fusion: development of improved deep CNN with atrous convolution-based remote sensing image fusion
Published in The Imaging Science Journal, 2023
S. Nagarathinam, A. Vasuki, K. Paramasivam
The training dataset is constructed in the proposed remote sensing image fusion model as it needs more amount of training data for the network. The actual satellite remote sensing images usually do not contain ground-truth images to make them as the target. Here, the ground-truth images of MS can be performed, and down-sampled PAN and MS images are considered as the input for the fusion network. On the other hand, it is very challenging to train the network with such an operation for three reasons. Firstly, the trained model contains poor fusion quality due to the simple colour texture in the remote sensing images as well as this model was not well performed to possess strong generalization ability. Secondly, the image resolution of remote sensing seems to be very low, along with less texture information, which makes it difficult to train the fusion model. Thirdly, it is cost inefficient for performing the remote sensing acquisition as it is necessary to have a high amount of training data. For solving these problems, the training dataset is built through the natural image set. Consider the image as the ground-truth images, and, the low resolution images correspond to the MS image. Finally, the enhanced PAN image is given as the input to the image fusion network.
Swell-shrink behaviour of an expansive soil stabilised by CCR-fly ash (FA) columns and CCR-FA blends
Published in Geomechanics and Geoengineering, 2022
Farahnaz Darikandeh, B. R. Phanikumar
For the repeatability of the digital image analysis, three samples were subjected to digital measurement analysis of 3-D shrinkage. The ImageJ software, which was used for this purpose, can measure the area through perimeter-defined image in pixels. A digital camera with image resolution of 4288 width, 2848 height and 300 pixel/inch, and JPG format with 3 Megabytes and ×1 digital zoom was used. After two months, equidistant markings were made with thread on the shrunk specimens (Figure 2(a)). The shrunk soil specimens in treated and untreated conditions after the first cycle and the fifth cycle were tested. The volumetric shrinkage strain (VS) was determined using the following equation developed by Puppala et al. (2004):
Determining the maximal inscribed rectangle of an irregularly shaped stone using machine vision
Published in International Journal of Computer Integrated Manufacturing, 2022
Yu-Ting Luo, Ching-Fang Chen, Syh-Shiuh Yeh
The main hardware of the proposed machine vision system comprises a backlight table, industrial camera and lens, and a laptop computer, as shown in Figure 1. With respect to the backlight table, the backlight is installed below the placement table in order to place the irregularly shaped stone to be detected. The industrial camera and lens are fixed above the backlight table to capture the stone image. This study uses a GigE DFK 23GM021 color industrial camera, and sets the captured image size as 1280 960. Moreover, this study uses an MDF010 lens with a focal length of 10 mm. The field of view of the image processing system is 300 225 mm2, and the working distance is 600 mm. The image resolution is 224 μm/pixel. The laptop computer executes the proposed image processing procedure and rotation histogram method to determine the maximal inscribed rectangle of the stone to be detected. The operating system of the laptop computer is Microsoft Windows 10, the central processing unit is Intel i7-4800MQCPU (3.7 GHz), and the graphic card is Nvidia Geforce GTX 870 M. The program development software used in this study is Microsoft Visual Studio 2015, and the programming language is C++. This study uses Open Source Computer Vision Library to perform image processing of the captured images.