Point target detection and subpixel position estimation in optical imagery

This paper addresses the issue of detecting point objects in a clutter background and estimating their position by image processing. We are interested in the specific context where the object signature significantly varies with its random subpixel location because of aliasing. Conventional matched filter neglects this phenomenon and causes consistent loss of detection performance. Thus, alternative detectors are proposed and numerical results show the improvement brought by approximate and generalized likelihood ratio tests in comparison with pixel matched filtering. We also study the performance of two types of subpixel position estimators. Finally, we put forward the major influence of sensor design on both estimation and point object detection performance.

💡 Research Summary

This paper tackles the classic yet still challenging problem of detecting isolated point‑like targets in optical imagery and estimating their locations with sub‑pixel accuracy. The authors begin by highlighting that conventional matched‑filter (MF) detectors assume a fixed target signature that is independent of the target’s exact position relative to the sensor’s pixel grid. In reality, because the imaging system samples a continuous point‑spread function (PSF) at discrete pixel locations, a target that falls anywhere within a pixel produces a digitally sampled signature that can vary dramatically—a phenomenon known as aliasing. When the sub‑pixel offset is ignored, the MF’s performance degrades, especially in low‑signal‑to‑noise‑ratio (SNR) regimes or when the pixel size is comparable to the PSF width.

To address this, the authors formulate a rigorous statistical model. The continuous scene containing a point target is convolved with the optical PSF, then sampled on a rectangular grid with spacing Δ. The resulting discrete image g