This paper focuses on the problem of low detection accuracy and low pixel image extraction accuracy of traditional small target detection and ship detection methods. An improved target detection algorithm based on constant false-alarm rate（ CFAR ）- dense conditional fandom fields （ DCRF）is proposed. The algorithm is based on the characteristics of small target and false alarm signal changes but different structural features. It uses the advantages of conditional fandom fields (CRF) multi-dimensional context (space, radiation) to achieve false alarm feature suppression, and introduces CFAR to improve the model and improve DCRF. Based on this model, experiments were performed under different conditions. The analysis results show that the algorithm can make full use of the global context information of the sea area, and can reduce the false alarm rate while maintaining a high detection rate.