Digital Image Processing of Remote-Sensed Data
Digital Image Classification
In remote sensing, the information hidden in multispectral image pixels or band pixels can be understood in a variety of ways, and image classification is one of those ways. The primary purpose of image classification is to enable computers to understand and categorize the content within images by assigning them labels or tags. This process helps machines recognize and classify the objects, scenes, or concepts present in an image, effectively providing a “digital explanation” of the image’s content. This information can be utilized for various purposes, such as land use and land cover mapping, agriculture monitoring, natural resource management, and environmental studies, among others. Image classification simplifies the data by categorizing the pixels into different classes and makes it easier for users to analyze and understand spatial patterns, trends, and relationships. In digital image processing, image classification is typically done before image segmentation, meaning that one would first identify the overall content of an image (classification) before dividing it into specific regions (segmentation) to analyze individual objects within the image with greater detail.
Types of Image Classification
Common classification procedures can be broken down into three broad techniques based on the method used: supervised (human-guided) classification, unsupervised (calculated by software) classification, and object-based image analysis.
Supervised Classification
In the supervised approach to classification, the image analyst supervises the pixel categorization process by specifying to the computer algorithm numerical descriptors of the various land cover types present in the scene. To do this, representative sample sites of known cover types, called training areas or training sites, are used to compile a numerical interpretation key that describes the spectral attributes for each feature type of interest. Each pixel in the data set is then compared numerically to each category in the interpretation key and labeled with the name of the category it looks most like.
Unsupervised Classification
Unsupervised classification is where the outcomes (groupings of pixels with common characteristics) are based on the software analysis of an image without the user providing sample classes. The computer uses techniques to determine which pixels are related and groups them into classes. The user can specify which algorithm the software will use and the desired number of output classes, but otherwise, it does not aid in the classification process.
Object-Based Image Classification
Supervised and unsupervised classification is pixel-based. In other words, it creates square pixels, and each pixel has a class. But object-based image classification groups pixels into representative vector shapes with size and geometry. It begins by considering each pixel as a separate object. Subsequently, adjacent pairs of image objects are merged to form bigger segments.
Pixel-Based Classification
Classification is done on a per pixel level, using only the spectral information available for that individual pixel (i.e. values of pixels within the locality are ignored). In this sense each pixel would represent a training example for a classification algorithm, and this training example would be in the form of an n-dimensional vector, where n was the number of spectral bands in the image data.
Machine Learning Classification
Machine learning classification is a supervised learning technique where algorithms learn to categorize data into predefined classes or categories. Essentially, it's about building a model that can predict which class a new data point belongs to, based on the patterns it learned from labeled data. Machine learning classification can be used in agriculture for classifying multiple crop types using multispectral satellite imagery and ground-truth yield data.
Deep Learning Classification
Deep learning classification involves training neural networks—especially convolutional neural networks (CNNs)—to recognize complex patterns and classify either individual pixels or entire image regions. CNNs are particularly well-suited for image processing tasks due to their ability to learn hierarchical features and spatial relationships within images.
Click on the following topics for more information on digital image processing of remote-sensed data.
