Although plenty of structural variant detecting approaches for human genomes can be looked up in the literatures, little has been acknowledged on the effectiveness of those structural variant softwares for plant genomes. It has been demonstrated frequent occurrences for those structural variant detecting softwares to find too many false structural variants. In this paper, we devote to detect deletions, insertions, inversions, in total of three kinds of structural variants occurring in Hedou12 genome in contrast to Williams82 genome. Aiming to enhance the precision of structural variant detections, we propose two new sequencing characteristic based probability models, which use the sequencing parameters of Hedou12 genome, as well as the parameters for Hedou12 paired-end reads to be aligned onto Williams82, to evaluate the probability a potential structural variant can occur in. To remove the false members from those potential structural variants, we propose a set cover problem model to describe formally on which potential structural variants it should accept to achieve as high as possible a probability summation. This will achieve a solution with more credible structural variants, which can be verified by comparing with DELLY and LUMPY.