Fuzzy target detection algorithm based on improved SSD and transfer learning

Fuzzy target detection as an important task to reflect the detection ability of underwater robot, the artificial target recognition based on the image taken by underwater robot has been widely concerned. However, there is no open standard fuzzy underwater image data set, and in the harsh deep-water fuzzy environment, it is difficult to collect large-scale marked underwater fuzzy optical images. At the same time, it is also hoped that the detection model has the ability to learn quickly from small samples in the case of as few samples as possible. Therefore, combining depth learning and transfer learning, a new method based on improved SSD and transfer learning is proposed. Firstly, we design a more accurate SSD network (underwater SSD) which is suitable for fuzzy underwater target detection. The features extracted from the detection network are highly representative. Secondly, we use the Transfer learning method to train the underwater SSD network, which can only use the tags in the air to identify fuzzy underwater objects, and have strong robustness in both the air and fuzzy underwater imaging modes. Finally, soft NMS is used to detect the target. The experimental results of the simulation data show that the algorithm not only overcomes the difficulties of the known data set of underwater target, but also effectively improves the accuracy of underwater target detection compared with the traditional deep learning method, reaching 82.31%, showing better detection performance.

DOA-Based Localization Method with Multiple Screening K-Means Clustering for Multiple Sources

The existing angle-based localization methods are mainly suitable for the single source. Actually, there often exists situation which contains multiple target sources. To solve the problem of localization of multitarget sources, this paper presents a K-means clustering method based on multiple screening, which can effectively realize the localization of multiple sources based on DOA (direction of arrival) parameters. The method firstly establishes a cost function of position coordinates by using DOA parameters from the measuring position coordinates and then solves the cost function to obtain a complete set of real position coordinates and fuzzy position coordinates. As the distribution of real target coordinates is concentrated and the fuzzy target positions are scattered, the K-means clustering method is adopted to classify the coordinate set. In order to improve the positioning accuracy, a multiscreening process is introduced to screen the input samples before each clustering, and it can be finally concluded that clustering centers are the position coordinates of the target sources. Meanwhile, the complexity analysis and performance verification of this method are proposed. Simulation experiments show that this method can efficiently realize ambiguity-free, highly precise localization of multitarget sources.

New Uncertainty Measure of Rough Fuzzy Sets and Entropy Weight Method for Fuzzy-Target Decision-Making Tables

In the rough fuzzy set theory, the rough degree is used to characterize the uncertainty of a fuzzy set, and the rough entropy of a knowledge is used to depict the roughness of a rough classification. Both of them are effective, but they are not accurate enough. In this paper, we propose a new rough entropy of a rough fuzzy set combining the rough degree with the rough entropy of a knowledge. Theoretical studies and examples show that the new rough entropy of a rough fuzzy set is suitable. As an application, we introduce it into a fuzzy-target decision-making table and establish a new method for evaluating the entropy weight of attributes.