Complete remote sensing digital image processing includes two parts: hardware system and software system. The satellite remote sensing data has a large storage capacity, requiring the coordination of large-capacity digital storage devices and software for storage and processing. Here, we mainly introduce the software processing part. The following is a professional image processing software interface, compared with commonly used office software, the various functions of the image processing system appear relatively scattered, and the connection between various menus is not close.
In a sense, the image processing system is more like an image processing toolbox. Depending on the image processing goals, users can call a single function or a combination of several functions, and not all processes are selected. Here, we summarize some typical processing functions and introduce the basic steps.
Data storage and management: storage management of different image data obtained by different sensors, read and display, and conversion output.
Image preprocessing: radiometric correction, geometric correction, image registration, atmospheric correction.
Image enhancement and conversion: image fusion, image enhancement processing, color synthesis, density segmentation, image cropping, and stitching uniform colors.
Classification and feature extraction: statistical analysis, feature extraction, image classification (supervised classification, unsupervised classification), thematic mapping, professional tools (multispectral, radar, terrain and other image processing).
Remote sensing images themselves have a large memory, and a Landsat remote sensing image with 1 scene and 7 bands has at least 200MB, while hyperspectral images may reach 1GB. Since entering the era of high-frequency production and accumulation of data in both time and space, remote sensing has also entered the era of big data, making remote sensing cloud services, storage management, rapid distribution and sharing trends more and more obvious.
Based on private cloud and hybrid cloud satellite remote sensing image digital storage, online update, management retrieval, and publication browsing have gradually become an inseparable important foundation for remote sensing data processing, and it will greatly enhance the efficiency of subsequent professional processing and business applications of remote sensing images.
Refers to the process of correcting or eliminating systematic and random radiation distortion or distortion caused by the acquisition and transmission system due to external factors, to eliminate or correct the image distortion caused by radiation error.
To put it simply, it is to remove the "noise" caused by the sensor or atmosphere, accurately reflect the ground conditions, improve the "fidelity" of the image, mainly to restore missing data, remove thin fog or prepare for stitching and change monitoring.
The role of radiometric correction in dynamic monitoring: In multi-temporal remote sensing images, in addition to changes in land use may cause changes in radiation values in images, the radiation values of unchanged land use in images at different times will also differ. If it is necessary to use the spectral information of multi-temporal remote sensing images to monitor the dynamic changes of land use, the radiation value differences of unchanged land use must be eliminated first.
Through relative radiometric correction, one image is used as a reference (or base) image, and the DN value of the other image is adjusted so that the same-named land use on the two temporal images has the same DN value. This process is also called spectral normalization of multi-temporal remote sensing images. This way, the changes in the radiation values of multi-temporal remote sensing images can be analyzed to achieve change monitoring, and thus complete the remote sensing dynamic monitoring of land use changes.
The random distribution map obtained by counting the number of pixels of each brightness in each image is the histogram of the image. Generally speaking, for images with a large number of pixels, the random distribution of brightness of pixels should be a normal distribution. If the histogram is non-normal distribution, it means that the brightness distribution of the image is too bright or too dark, or the brightness is too concentrated, and the image contrast is small. It is necessary to adjust the histogram to a normal distribution to improve the image quality and facilitate the identification of the outlines of the land and the extraction of information.
The features of the target on the remote sensing image are the reflection of the difference in the radiation of the electromagnetic wave of the target on the remote sensing image. Based on the features of the satellite remote sensing image, the process of identifying the properties, spatial position, shape, size, and other attributes of the target is called remote sensing information extraction.