Target area segmentation method and device in ground image

Abstract

The invention discloses a target area segmentation method and a device in a ground image. The method comprises the following steps: for at least one ground image in a test image data set, inputting the ground image into a pre-trained deep full convolutional neural network model to obtain a probability matrix; wherein each element of the probability matrix represents the probability that a pixel point corresponding to the element in the ground image belongs to at least one type of target area; wherein the deep full convolutional neural network model is obtained by training different types of target regions of a plurality of ground sample images in an image sample set; according to the probability matrix, obtaining a pre-selected region of each type of target region included in the ground image; and processing the pre-selected area to obtain a segmentation result of each type of target area of the ground image. According to the method, each pixel of the ground image is classified throughthe deep full convolutional neural network, accurate segmentation of different target areas is realized, the realization process is simple, and the method is not easily influenced by the imaging quality.

Description

technical field [0001] The invention relates to the field of computer vision, in particular to a method and device for segmenting a target area in a ground image. Background technique [0002] Ground images refer to films or photos that record the size of electromagnetic waves of various ground features, including remote sensing images (Remote Sensing Images) and high-altitude overlooking images. Ground image segmentation refers to the technology and process of dividing the ground image into target areas with different characteristics and extracting the target of interest. Here, the feature can be the grayscale, color, texture of the pixel, etc. The pre-defined target can correspond to a single area, or Can correspond to multiple regions. Ground image segmentation is a key step from ground image processing to ground image analysis, and occupies an important position in image engineering. [0003] In the prior art, the target area of ​​the ground image is segmented, general...

AI Technical Summary

Problems solved by technology

[0003] In the prior art, the target area of ​​the ground image is segmented, generally using a supervised learning model such as a support vector machine (SupportVector Machine, SVM). The requirements are high, and the ground image needs to be transformed three-dimensionally or other information sources are required, and only the pixels of the ground image can be classified into two categories. Each operation can only obtain the segmentation result of a specific target area of ​​the ground image. Cannot simultaneously distinguish and segment various target regions in ground images

It can be seen that the existing ground image segmentation has a complex implementation process, is easily affected by the imaging quality, and cannot quickly and accurately obtain the segmentation results of different target areas.

Images