In the applications of optical fiber sensor array, many aspects of signal processing are affected by the error of array shape. For example, beam-forming and array shape are closely related. To reduce the effects of shape deformation, an array shape estimation method based on polynomial curve fitting algorithm is proposed in this paper. In the simulation, the heading sensors are placed in the fiber sensor array to obtain the angle information on certain points of the array. With polynomial curve fitted, the estimated function of angle and array length is built. According to the geometry relationship, the X- and Y-axis data of all the points we need on the array are calculated. At last, we compare the results with real array shape in the model, and demonstrate the efficiency of the method. In addition, beam forming simulation is used to check the method. Beam-forming performance of the fiber sensor array system is improved.
This paper attempts to develop an unsupervised learning approach for airplane detection in remote sensing images. This novel airplane detection method is based on circle-frequency filter and cluster-based co-saliency detection. Firstly, the CF-filter method is utilized as the coarse detection to detect target airplanes with some false alarms. Then, we collect all the detected targets and use cluster-based co-saliency detection to enhance the real airplanes and weaken the false alarms, so that most of the false alarms can be eliminated. Experimental results on real remote sensing images demonstrate the efficiency and accuracy of the proposed method.