Research And Implementation Of Port Segmentation Algorithm In Remote Sensing Image

Remote sensing image object segmentation, pattern recognition and image analysis is an important step. Remote sensing image segmentation has a very strong objective characteristics and image type specific, and therefore the target segmentation algorithm must be carefully designed in order to achieve better image segmentation results.Port as an important military base and civil construction, research information about it, is of great strategic significance and civil significance. Port object segmentation and recognition technology has important military applications. Remote sensing image segmentation and research objectives of the harbor, in the modern era, has been in the field, whether civilian or military field tasks are eager to become a hot research topic in the field of image analysis. Port of remote sensing image segmentation and extraction of research, its complexity and the diversity of characteristics of the environment, to achieve accurate image segmentation results, you need to select the appropriate method and specific algorithms for specific target.The main content of this paper is to study and implementation of remote sensing image segmentation algorithm in the port of destination, remote sensing images intercept from google earth. Select two remote sensing images, analyzing the characteristics of the port of destination information: gray features and structural features. On this basis, select the threshold segmentation method and pulse coupled neural network segmentation method for image segmentation.The main contents are as follows:1.Introduction of the image threshold segmentation method, based on the inter-dimensional Otsu on the study of two-dimensional Otsu between determine the optimal threshold for image segmentation, segmentation harbor achieve goals.2. A detailed study of an improved image thresholding method, the weighting parameter intuitionistic fuzzy entropy thresholding algorithm. Increasing the weight parameter inintuitionistic fuzzy entropy segmentation method on the basis of improved intuitionistic fuzzy sets lack of fuzzy uncertainty measure limitations. The purpose is to improve the effectiveness of traditional thresholding segmentation method, and to a certain extent, to eliminate or suppress the influence of factors on cloud segmentation results.3.Considering the threshold method own limitations, one or a set of thresholds can not be accurate segmentation of remote sensing images in the port target, based on this study the segmentation algorithm based on neural networks: pulse coupled neural network(PCNN-Pulse Coupled Neural Networks) segmentation algorithm is introduced and split core principle of this algorithm, taking into account the inadequacies of the algorithm itself, adding edge algorithm and denoising step of segmentation optimization.Several segmentation algorithm to achieve the above, through the analysis and discussion of the segmentation results, verify their feasibility. The results show that these algorithms to some extent, improve the efficiency and to optimize the segmentation segmentation result.