Infrared False Alarm Detection Based On Fractal And Morphological Filtering

In earth observation systems,especially in the detection of small targets,the detection and identification of long-range infrared targets has an important role in the military and civilian fields.However,there are a large number of high-radiation areas on the earth's surface,which have become false alarm sources for target detection,causing interference to target detection,resulting in a high target false detection rate and interfering with military early warning systems.In order to improve the system performance and small target detection accuracy,the method proposed in this thesis uses suppression of false alarm sources as an auxiliary means for small target detection.The texture analysis of the false alarm source found that the fractal can well represent the false alarm source,and at the same time,specific types of signal components in the infrared image can be efficiently characterized by the same type of dictionary,and morphological filtering can remove irrelevant information and enhance useful information.Therefore,this subject proposes a method combining fractal features and morphological filtering and a method combining fractal dictionary and double-structure morphological index for the detection of false alarm sources in infrared imaging.The research work in this thesis is as follows:(1)The extraction and analysis of fractal characteristics of false alarm source in infrared imaging were studied.The texture of false alarm source is found to have self-similarity,that is,it has fractal features,so the false alarm source is enhanced by extracting the fractal features of false alarm source.The fractal features extracted in this project include fractal dimension,fractal area and fractal features of multifractal analysis.The algorithm simulation test found that the fractal area extraction effect is the best,and the method has the strongest background suppression effect.(2)The algorithm of false alarm source detection based on Fractal dictionary learning is studied.The proposed method is based on the fractal feature of the false alarm source,and the specific type of signal components in the infrared image can be efficiently represented by the same type of dictionary.This method has achieved certain results in false alarm source detection,and it has better performance in quality indicators such as ROC curve,PR curve,AUC,AUCpr,F-measure and IOU.(3)The method of morphological filtering was researched,and based onmorphological reconstruction,double operator morphological filtering and morphological building index,the morphological index of double structure was proposed.The proposition of this index makes the background suppression effect stronger and makes a huge contribution to the subsequent combination of fractals.(4)The detection of false alarm sources by combining fractal and morphology is studied.Based on the previous basis,a false alarm source detection algorithm combining fractal features and morphological filtering,and a false alarm source detection algorithm combining fractal dictionary and dual-structure morphological index are proposed.Through simulation tests,it is found that the algorithm combining the fractal dictionary and the double structure morphology index performs better on the ROC curve and the PR curve,and has a higher accuracy rate under the same recall rate.Its F-measure value and IOU value are greater than other algorithms,which shows that it has better detection effect.