Emissivity domain thermal infrared hyperspectral anomaly detection method based on partitioning and low-rank prior

Abstract

The invention relates to an emissivity domain thermal infrared hyperspectral anomaly detection method based on partitioning and low-rank prior. Firstly, temperature and emissivity inversion is performed on an original image to obtain an emissivity graph and a temperature graph of the image, and then the emissivity graph is segmented into a plurality of homogeneous regions by using the temperature graph and the radiance graph in combination with a Popts-based image segmentation algorithm; according to observation, in a local uniform region, the background signal has enhanced low-rank performance, and the anomaly is represented as spatial sparsity; based on this observation, background pixels may be low rank reconstructed from a set of basic background signals, while anomalies may be represented with sparse residuals; and then low-rank sparse matrix decomposition is carried out on the original hyperspectral data matrix by utilizing the extracted background end member, and a part of noise is removed and background information which is purer than that of the original image can be acquired; finally, the spectral difference between the anomaly and the background is calculated by combining the Mahalanobis distance with the original emissivity image and the background information so as to realize anomaly and background separation.

Description

Technical field[0001]The present invention is based on the field of remote sensing image technology, and in particular, the present invention relates to a method of thermal infrared high spectroscopy abnormal detection based on blocking and low rank priority.Background technique[0002]High spectroscopy allows simultaneous observation of the geometric and physical properties of the material, which makes accurate distinguishes different goals. Target detection is one of the important research fields of high-spectral information processing. High spectroscopy target detection is generally divided into an abnormal detection and features based on feature detection. Among the abnormal detection tasks, there is no prior knowledge about the abnormal or background, which is an observation value that has a significant difference in the spectral characteristics of adjacent backgrounds. The radiation of the long-wave infrared spectrum zone measured by the sensor mainly comes from the surface obje...

AI Technical Summary

Problems solved by technology

Second, traditional statistical methods are difficult to accurately describe the complex background of hyperspectral images

Third, because the thermal infrared sensor is very sensitive, the signal-to-noise ratio of the thermal infrared image is relatively low, and the noise can easily cover up the spectral characteristics of the ground object. In addition, the emissivity spectrum of the object has a small change range, so the distance between the target and the background is relatively low. The spectral contrast of the

Due to the existence of the above-mentioned problems, the anomaly detection performance of thermal infrared hyperspectral images is often poor.

Images