Research On Background-Clutter-Based Target Acquisition Performance Model

The target acquisition performance of an electro-optical imaging system is of great importance. With the improvement in the design of optical system, detector and circuit, the effect of the imaging system on the target acquisition performance is weakened, which makes the target background characteristics become more and more important. Therefore, a correct understanding and a more suitable quantification of the target background characteristics is the key to establish a more perfect and accurate performance prediction model. A novel clutter metric based on the embedded hidden Markov model(EHMM) is proposed and used to modify the target acquisition performance model in this paper.Considering the 2-D characteristics of the target image and the background image, a two-dimensional mathematical model-- embedded hidden Markov model is selected to characterize the image features. The clutter value changes when different targets are considered, thus the feature extraction parameters are set adaptively according to the target. The process of training optimization model parameters is consistent with the physical mechanisms of an observer to detect and record the target, then the possible state path and the occurrence probability(a parameter that can quantify the similarity between the model and the observation sequence) to generate the observation sequence of target and background are recorded through decoding. And the similarity between the target and the background is calculated through statistical analysis. The whole process focuses on the similarity between the target and the background, the most import feature of background clutter. And it consists with the process that the detection path is always optimized when an observer detects the target. Therefore, compared to the clutter metrics existing, the embedded-hidden-Markov-model based clutter metric(EHMMC) proposed in this paper agrees with subjective test better.The target acquisition performance model under complex background conditions is modified by the newly proposed EHMM based clutter metric. The local detection probability model for complex background and the target-like detection probability model based on EHMMC are established. And experiment results show that the modified model is better in the performance prediction of an electro-optical imaging system.