Landmind detection with IR sensors using Karhunen Loeve transformation and watershed segmentation

In this paper, we present our idea of using the Karhunen Loeve transformation (KLT) and watershed segmentation to detect landmine objects from infrared images. On doing this, we proposed a simplified process for reducing the computation in the Karhunen Loeve transformation using a smaller number of images than traditional methods do. We effectively used the marker based watershed segmentation to detect the mines with high performance detection rate. We tested our proposed method on three different mine fields with two different soil types. Our proposed method consists of four stages: feature extraction, enhancement, object segmentation, and object recognition. The results are promising.