A new algorithm for small target detection in liquid image sequence

Abstract

Small target detection is a crucial problem for impurity detection in liquid. In order to get better detection effectiveness, a new algorithm based on multi-threshold classification and forward verification is described in this article. This algorithm can detect impurity-one kind of small target in liquid, which is able to detect impurity with full intelligent. After background suppression, an adaptive multi-threshold classification is adopted for the extraction of potential targets. A time pipeline is set up for energy accumulation. After forward searching, the potential target is assessed to the real target by a reasonable threshold. The S/N sensitivity of the algorithm is greatly improved. Some experiments using real ampoule liquid have given satisfying results.