Interest of correlation-based automatic target recognition in underwater optical images: theoretical justification and first results

摘要: In this paper, we explore the use of optical correlation-based recognition to identify and position underwater man-made objects (e.g. mines). Correlation techniques can be defi ned as a simple comparison between an observed image (image recognize) reference image; they achieved extremely fast. The result is more or less intense correlation peak, depending on resemblance degree coming from database. However, perform good decision, should compare our with huge database references, covering all appearances search. Introducing influence speed and/or quality. To overcome limitation, propose composite filter techniques, which allow fusion several references drastically reduce number needed comparisons images. These recent have not yet been exploited in context. addition, for integrating some preprocessing directly manufacturing step enhance visibility objects. Applying one reduces processing by avoiding unnecessary Fourier transforms their inverse operation. We want obtain fi lters that are independent noises contrast problems found videos. achieve create containing scales viewpoints, 3D computer-generated