The Design And Research Of The Wave Monitoring System

China is a big maritime country, which has numerous islands and long coastline. So the wave monitoring has an extremely important strategic and practical significance to China. Establishing a visual remote monitoring system is an effective means for monitoring the waves. In this paper, combined with the specific project background, a wave monitoring system is put forward in the view of the key technology of moving object detection based on the video. The research contents of this paper are summarized as follows (the main innovation points are embodied in 2 and 3):1) Background motion compensation. To solve the problem of background motion caused by camera movement, a simplified ORB feature detection and matching algorithm is used to match two images. Then, the PROSAC algorithm is used to fit model parameters, obtaining the affine matrix between two images. Finally, background motion compensation is applied to the image frames based on the affine matrix.2) Threshold segmentation. According to the need to retain the useful information completely when the gray image is converted into binarization map and have a good anti-noise performance, an adaptive binarization method based on weighted integral image is presented in this paperc based on the grid-based model and the weight-based model. Experiment proves that this algorithm shows good robustness in the complex wave environment.3) Moting wave detection. In the view of the problem that the specific wave detection cannot satisfy the specific wave, this paper presents a wave edge detection algorithm based on mixed differential method. First of all, the three frame differential method is used for rapid classification of scene areas, dividing out the movement area. Then, the morphologic processing is implemented: dilation and erosion. Subsequently, the Sobel operator is used to extract moving object edge. Finally, the interference movement is removed by using the method of background subtraction based on gaussian mixture model. Experiments show that the algorithm has good robustness to the complex wave scene, and it also can extract the edge of moving wave accurately while keeping high efficiency.4) Benchmark extraction. In order to extract benchmark as a yardstick, a clear initial image is selected and the benchmark is extracted using the threshold segmentation algorithm based on HSV space in the initial image in this paper. Then, the initial image and the current image are matched. Finally, the position of benchmarking in the current image is calculated by the affine matrix.Moreover, in the view of the problem of obtaining the historical video shooting time, a probability model of character recognition algorithm based on the Bayes classifier and the principle of similar shapes is developed.Based on the research above, a complete wave monitoring system is built in this paper, and the actual video uses to experimental analysis of the algorithms. The experimental results fully prove the feasibility and effectiveness of the algorithm proposed in this paper.