摘要：Image analysis is an important means of the the nuclear environment monitoring. In this paper, a novel moving target detection called Local Trinity Patterns which is based on Local Binary Patterns algorithm is presented, Based on The standard LBP proposed by Ojala3, it mainly captures the texture information, and in some circumstances it results in misidentification. The proposed LTP feature, in contrast, captures the gradient information and some texture information. Moreover, the proposed LTP are easy to implement and computationally efficient, which is desirable for real-time applications. Experiments show that the algorithm is effective, can significantly improve the nuclear environment detection performance and produce state of the art performance.
