Research Of The Jitter's Effect On The Detection And Tracking Of Point Targets

For infrared-optical imaging guidance systems, if the distance between theinfrared image sensor and a target is long, the image of the target is point-like.Multi-frame signal accumulation is often used to improve the signal-to-noise ratio(SNR) and the detection probability of infrared-optical systems. Existing theories andalgorithms of energy accumulation are made under the hypothesis that the positions ofthe target in multi-frame are aligned to each other. However, there is randominter-frame motion caused by jitter of the target's image. So the hypothesis onalignment is not tenable, especially for point-like targets.In this dissertation a mathematical model is set up for describing the energyaccumulation process of a jittering point target. Then, the dissertation is dedicateditself to research the effect of multi-frame energy accumulation on SNR and on thedetection probability. After that, the technique for the design of a filter to suppressjitter and to realize effective energy accumulation is proposed. The main innovationsof the dissertation can be listed as follows:1. Generalized jitter is defined. While jitter has different definitions in differentfields, the common result of jitter is creating the inter-frame random motion of pointtargets. From the point of view of point spread function (PSF), jitter is defined astime-dependent random displacement of center position of the system's PSF. Thedefinition endows jitter with specific mathematical model.2. A random accumulation model (RAM) is proposed that aims at handling pointtargets in jitter. Random jitter makes it difficult to improve SNR by energyaccumulation. It is demanded to set up a mathematical model to describe the processof random accumulation and to analyze the effect of jitter on detection performance ofpoint targets. This would be important for the design of detection-tracking systems of point targets.3. The RAM is used to analysis the space-distribution characteristics ofshort-exposure speckle images and long-exposure speckle images of point targets withturbulent effects. It is manifested that the long-exposure speckle function is thecross-correlation of the short-exposure speckle function and the jitter distributionfunction. The atmospheric turbulence randomly interferes the transmission of the lightwaves came from the objects. This causes a location displacement of the image on thefocal plane, creates a blur of the image, and makes the SNR very low. The image ofpoint target through atmospheric turbulence is exhibited as a speckle image. Speckleimages are used as the fundamental elements to search the statistical characteristics ofthe turbulent effect. By using such a novel idea for the analysis of thespace-distribution of long-exposure speckle images, meaningful conclusions can beachieved.4. A preprocessing algorithm of point target detection based on temporal-spatialbilateral filter using adaptive neighborhoods is proposed. The algorithm combines theadvantages of both bilateral filtering and adaptive neighborhood. It can remove noiseas well as enhance the contrast of the point target against its background efficientlyand rapidly. The algorithm has not a strict restrict on the beeline motion of the target.As a result, it is more amenable to handle jittering point targets.5. Dynamic displacement field (DDF) theory is used to analysis jittering pointtarget, and a position displacement filter based on dynamic displacement field theoryis proposed. From the theory of DDF model, jitter is a kind of high-frequencyvariation of image displacements. Naturally, the filtering of displacements in thefrequency domain should suppress jitter. The position displacement filter is apoison-time sequence filter. The filter can remove high-frequency jitter effectivelyand keeping low-frequency movement at the same time.