China
Africa
Explore chapters and articles related to this topic
Advanced Optical Observations
Published in Victor Raizer, Optical Remote Sensing of Ocean Hydrodynamics, 2019
In general, any ocean optical data of high resolution are much better than data with low resolution. As it follows from theoretical and empirical studies, spectral dependences of the reflection coefficient of light (spectral reflectance) for free pure water surface in the range of 0.45–0.90 μm is <5% at nadir (Chapter 3), i.e., the difference within IKONOS multispectral bands is small. It is also known from aerial experiments that sea surface reflectance is color neutral and modulations of the light by wave slopes are very weak. It means, in particular, that spectral processing is more efficient for PAN data with 1-m resolution than for any MS band data with 4-m resolution because of PAN data have a better quality and highest pixels intensity. If necessary, PAN data can be integrated into low resolution data and in this case, they will be also better than multispectral data. However, fusion between individual multispectral bands (Blue, Green, Red, NIR) and PAN band makes sense to investigate in terms of spectral characterization of ocean optical data. This operation known as “pansharpening” is often used in multispectral imagery for land cover classification (Alparone et al. 2015; Pohl and van Genderen 2017). Although fusion methods do not necessarily improve visual quality of ocean PAN images, in our case, pansharpening provides much better detalization and specification of generated digital 2D FFT spectra (Figure 7.15). Fused spectra are more informative than non-fused spectra.
Applications of Computer Vision
Published in Manas Kamal Bhuyan, Computer Vision and Image Processing, 2019
In remote sensing applications, it is always desirable to have high spatial resolution and narrow spectral bandwidth images which are obtained from a sensor. Panchromatic images have high resolution, while multi-spectral images have low spatial resolution. Hence, a multi-spectral image with high spatial resolution gives better visual perception. For this, pansharpening is performed to obtain high spatial resolution multi-spectral images. Figure 5.47 shows one example of pansharpening using multi-spectral and panchromatic images. For performing image fusion, the source images are obtained from [119,154,183,214].
Singular value decomposition and saliency - map based image fusion for visible and infrared images
Published in International Journal of Image and Data Fusion, 2022
C. Rajakumar, S. Satheeskumaran
A fusion of high spatial resolution panchromatic image and low-resolution multispectral image is known as pan sharpening which is applied to generate high spatial resolution multi-spectral images in remote sensing (Yin 2015). Fusion methods in remote sensing satellite images can be classified into three categories namely component substitution methods (Choi et al. 2010), multi-resolution analysis-based methods (Garguet-Duport et al. 1996) and degradation model-based methods (Junli et al. 2005). There are two groups to fuse images namely spatial and frequency domain. Spatial domain methods deal with the manipulation of pixels directly. Few spatial domain techniques are averaging, principal component analysis (Desale and Verma 2013) and high pass filtering. Frequency domain methods are based on stationary wavelet transform (Borwonwatanadelok et al. 2009), discrete cosine transform (Chu and Zhu 2006), wavelet transform (Pajares and De La Cruz 2004) and contourlet transform (Yang et al. 2007) etc. in which frequency components have been altered and it provides a remarkable improvement in spatial and spectral characteristic compared to spatial domain methods. Sparse, dictionary and lower rank approximation has been received great attention in image fusion. Dictionary-based method (Li et al. 2013b) in which source images are decomposed into patches based on geometric direction. These patches are sparsely coded by sub dictionaries via an online dictionary learning algorithm. Max fusion that rule combines all sparse and sub-dictionary constructs the fused image.
A context-driven pansharpening method using superpixel based texture analysis
Published in International Journal of Image and Data Fusion, 2021
Hind Hallabia, Habib Hamam, Ahmed Ben Hamida
Over the last two decades, a large number of pansharpening approaches have been proposed, which can be classified into three major families (Vivone et al. 2018b, Meng et al. 2019): Component Substitution (CS), Multi-Resolution Analysis (MRA) and Model-based-Methods (MMs). The CS-based methods consist first of transforming the low-resolution MS image space into a new colour space. Then, the intensity/first component is substituted totally or partially with the histogram-matched PAN image. Finally, inverse transform is applied to generate the high-resolution MS image. Representative approaches belonging to this category are Intensity Hue and Saturation (IHS) (Carper et al. 1990, Tu et al. 2001), Principal Components Analysis (PCA) (Chavez et al. 1991), and Gram-Schmidt transformation (GS) (Laben and Brower 2000). These methods present relatively good performances in terms of spatial quality, but they may suffer from spectral distortions (Xu et al. 2014). To tackle this issue, adaptive CS methods, such as Gram-Schmidt Adaptive (GSA) (Aiazzi et al. 2007), Band-Dependant Spatial Detail (BDSD) (Garzelli et al. 2008), and Partial Replacement Adaptive CS (PRACS) (Choi et al. 2011), have been proposed, taking into account adaptive weights and injection gains.
Varying weighted spatial quality assessment for high resolution satellite image pan-sharpening
Published in International Journal of Image and Data Fusion, 2022
Soroosh Mehravar, Farzaneh Dadrass Javan, Farhad Samadzadegan, Ahmad Toosi, Armin Moghimi, Reza Khatami, Alfred Stein
In the proposed SQA section of this study, we aimed to address this issue by proposing an efficient SQA procedure that can help the RS community with the purpose-oriented estimation of the spatial quality for pan-sharpened images. Using the proposed spatial quality assessment method along with accurate spectral quality assessment methods, the superior methods used for enhancing (pan-sharpening) the images can be reliably determined. After any image enhancement process and before the mapping projects (i.e. urban mapping, tree species mapping), the enhanced images should be spatially and spectrally assessed. The proposed method helps in finding the best image enhancement methods (i.e. pan-sharpening methods) that are suitable for mapping purposes.