Explain image classification techniques in remote sensing.

Image classification assigns pixels to land cover classes based on spectral characteristics. Supervised classification: analyst identifies training samples for each class, algorithm (maximum likelihood, SVM, random forest) classifies remaining pixels based on spectral signatures. Unsupervised classification: algorithm (ISODATA, K-means) groups pixels into clusters based on spectral similarity, analyst labels clusters. Accuracy assessment uses confusion matrix comparing classified image to ground truth. Object-based classification considers spatial context and texture. Deep learning methods increasingly used for complex classification.