Sparse-Representation-Based Clutter Metric

With the development of electro-optical imaging technology, background clutter has become an important factor affecting target acquisition performance of imaging system. It is becoming more and more important to measure the affection of clutter accurately for building a perfect target acquisition performance model.Based on the analysis of the definition and the physical essence of clutter, its development and the target-searching characteristic of human vision, a novel clutter metric called Sparse-Representation-Based clutter metric is proposed by combining theory of sparse representation with the perception of human vision in target detection. With the use of the relevance analysis and the error analysis, the Search₂ database is used to explain the feasibility of the SRC metric. What's more, compressed sensing theory is studied followed by the primary proposition of another novel clutter metric called Compressed-Sensing-Based clutter metric.The clutter metrics proposed in this paper can simulate the characteristic of the process of feature extraction and target researching of human vision reasonably. Besides, they are based on the characteristic of the background and the target, so they can match up to the physical significance of the clutter. Further more, the experience results show that their prediction results correlate well with the detection probability obtained by subjective experiments for human visual systems.