SAR (Synthetic Aperture Radar) image based automatic water area detection method

Abstract

The invention provides an SAR (Synthetic Aperture Radar) image based automatic water area detection method. The SAR image based automatic water area detection method comprises the following steps of a first step, performing de-noising processing on an SAR image; a second step, computing the gradient of an input image through an edge detection operator and obtaining a gradient image of the SAR image; a third step, improving the contrast ratio of the SAR image through a contrast ratio stretching conversion algorithm; a fourth step, achieving secondary self-adaption threshold segmentation through a double-peak iterative method and obtaining two binary image; a fifth step, performing post processing on the binary image obtained in the fourth step through morphology and obtaining a crude extracting target candidate area; a sixth step, removing false alarm, marking the target candidate area extracted from the fifth step, performing statistics of the area of every connecting dark spot, setting an area threshold value according to the resolution ratio of the SAR image, removing the dark spots with the area less than the preset area and marking extracted water areas in an original drawing.

Description

technical field [0001] The invention relates to a water area automatic detection method based on SAR images, which belongs to the technical field of image processing. Background technique [0002] Due to the characteristics of all-day and all-weather, Synthetic Aperture Radar (SAR) has been widely used in military, agricultural, aviation and other fields. Among them, water area detection is an important aspect of SAR image application. [0003] Generally, the surface of water body on land is relatively uniform and smooth. The electromagnetic waves emitted by SAR onto the surface of water body are equivalent to specular reflection, so the backscattering coefficient of water body is small, and the water area presents a single and uniform black dark spot area with single texture feature in the SAR image. . The surface of land, mountains, buildings and other objects is uneven, and electromagnetic waves will reflect slowly on the surface, so these objects have a large backscatt...

AI Technical Summary

Problems solved by technology

Among them, the extraction methods based on texture features mainly include algorithms based on gray level co-occurrence matrix, algorithms based on transformation, algorithms based on shape features, etc. This method is generally more complicated to calculate. For example, the algorithm based on gray level co-occurrence matrix needs to extract each The gray level co-occurrence matrix of pixels, the calculation process is very time-consuming

[0005] The method based on morphological extraction mainly uses erosion, dilation and their combined operations to extract image components useful for expressing and describing the shape of the region, such as boundaries, skeletons, convex hulls, etc. Processing or post-processing, such as filtering, thinning, pruning, etc., but the size of the structural elements and the number of operations of this method are difficult to determine

[0006] The method based on support vector machine (SVM) first needs to determine which texture features to choose and calculate these texture features, and then use SVM to train and classify, and finally get the extracted target, but it is difficult to determine the optimal combination of texture features, and the amount of calculation is also very large. Big

[0007] The method based on region growing is to determine the rules of region growing according to the target imaging characteristics and prior knowledge, but choosing a suitable growing rule not only needs a lot of experimental training, but also is sometimes very difficult.

[0008] The common disadvantages of the above methods are large amount of calculation, slow speed, difficult to meet real-time requirements, and need to use prior knowledge and a large number of experimental training to determine some features or rules, sometimes the determination of these features or rules is very difficult

Images