Simulation method and system for ship target in Gaofen-4 multispectral remote sensing image based on Poisson fusion and histogram matching

Abstract

The invention discloses a simulation method and system for a ship target in a Gaofen-4 multispectral remote sensing image based on Poisson fusion and histogram matching, and relates to a simulation method and system for a ship target in a remote sensing image. The invention aims to solve the problem that the ship target type in the Gaofen-4 remote sensing satellite image cannot be accurately identified by the existing method. The method comprises the following steps of: 1, cutting an interested five-waveband target image from a Gaofen-2 multispectral remote sensing image, and determining a target type of the cut target image; 2, respectively carrying out down-sampling on the cut five wave band target images to obtain five wave band images after Gaofen-2 down-sampling; 3, selecting and cutting a five-waveband image of a reference target from the Gaofen-4 multispectral remote sensing image to be added with the simulation target; 4, obtaining a processed high-resolution No.2 multispectral target map; and 5, obtaining a Gaofen-4 multispectral image added with the simulation target. The method and system are applied to the field of target simulation of remote sensing images.

Description

technical field

[0001] The invention relates to a ship target simulation method and system for remote sensing images. Background technique

[0002] When detecting sea surface ship targets on Gaofen-4 remote sensing satellite images, the type of ship targets cannot be confirmed usually due to the limitation of image spatial resolution. In addition, the number of ship targets in the images is usually small, resulting in It cannot provide effective data support for target detection. In order to overcome the above difficulties, researchers need to simulate the ship targets in remote sensing images to meet the research needs.

[0003] In order to provide an effective and sufficient number of ship targets, the method of target simulation in remote sensing images has gradually attracted the attention of domestic scholars. The characteristics of the simulated target should be as close as possible to the real target. There are two existing methods. One is to use 3D modeling softwa...

Examples

specific Embodiment approach 1

[0022] Specific implementation mode one: combine figure 1 To illustrate the implementation mode, the specific process of the simulation method of the ship target in the Gaofen-4 multi-spectral remote sensing image based on Poisson fusion + histogram matching in this embodiment is as follows:

[0023] Step 1. Cut out the five-band target image of interest from the multi-spectral remote sensing image of Gaofen-2 and determine the target type of the cropped target image (in view of the sub-meter high resolution of Gaofen-2, it can be manually interpreted method to determine its target type);

[0024] Step 2. Down-sampling the five-band target images cut out in Step 1 respectively, to obtain the five-band images after Gaofen-2 downsampling;

[0025] Step 3. Select and crop a five-band image of a reference target from the Gaofen-4 multi-spectral remote sensing image to be added to the simulated target;

[0026] The size of the reference target should be as close as possible to th...

specific Embodiment approach 2

[0029] Specific embodiment 2: The difference between this embodiment and specific embodiment 1 is that in the step 2, the five band target images cut out in the step 1 are respectively down-sampled, and the five band target images after the down-sampling of Gaofen No. 2 are obtained. Band image; the specific process is:

[0030] Down-sample the five band target images cut out in step 1 to a spatial resolution of 50m (the spatial resolution of the Gaofen-4 remote sensing image is 50m), and obtain the five down-sampled images of Gaofen-2. band image.

[0031] Other steps and parameters are the same as those in Embodiment 1.

specific Embodiment approach 3

[0032] Specific embodiment three: the difference between this embodiment and specific embodiment one or two is that in the step five, the position of the simulated target is selected as the area of ​​the target to be added in the Gaofen No. 4 multispectral image of the simulated target to be added, This area should meet the objective conditions for the existence of the target (for example, the ship can no longer be on land), and the Poisson fusion algorithm is used to add the multi-spectral target map of Gaofen-2 processed in step 4 to the simulated target selected in the multi-spectral image of Gaofen-4. position, to obtain the multi-spectral image of Gaofen-4 with the added simulation target; the specific process is as follows:

[0033] Poisson fusion is an image stitching method based on the image gradient field. The GF-2 multi-spectral target image processed in step 4 and the GF-4 multi-spectral image to be added to the simulation target are divided into five according to t...

technical field

[0001] The invention relates to a ship target simulation method and system for remote sensing images. Background technique

[0002] When detecting sea surface ship targets on Gaofen-4 remote sensing satellite images, the type of ship targets cannot be confirmed usually due to the limitation of image spatial resolution. In addition, the number of ship targets in the images is usually small, resulting in It cannot provide effective data support for target detection. In order to overcome the above difficulties, researchers need to simulate the ship targets in remote sensing images to meet the research needs.

[0003] In order to provide an effective and sufficient number of ship targets, the method of target simulation in remote sensing images has gradually attracted the attention of domestic scholars. The characteristics of the simulated target should be as close as possible to the real target. There are two existing methods. One is to use 3D modeling softwa...

Examples

specific Embodiment approach 1

[0022] Specific implementation mode one: combine figure 1 To illustrate the implementation mode, the specific process of the simulation method of the ship target in the Gaofen-4 multi-spectral remote sensing image based on Poisson fusion + histogram matching in this embodiment is as follows:

[0023] Step 1. Cut out the five-band target image of interest from the multi-spectral remote sensing image of Gaofen-2 and determine the target type of the cropped target image (in view of the sub-meter high resolution of Gaofen-2, it can be manually interpreted method to determine its target type);

[0024] Step 2. Down-sampling the five-band target images cut out in Step 1 respectively, to obtain the five-band images after Gaofen-2 downsampling;

[0025] Step 3. Select and crop a five-band image of a reference target from the Gaofen-4 multi-spectral remote sensing image to be added to the simulated target;

[0026] The size of the reference target should be as close as possible to th...

specific Embodiment approach 2

[0029] Specific embodiment 2: The difference between this embodiment and specific embodiment 1 is that in the step 2, the five band target images cut out in the step 1 are respectively down-sampled, and the five band target images after the down-sampling of Gaofen No. 2 are obtained. Band image; the specific process is:

[0030] Down-sample the five band target images cut out in step 1 to a spatial resolution of 50m (the spatial resolution of the Gaofen-4 remote sensing image is 50m), and obtain the five down-sampled images of Gaofen-2. band image.

[0031] Other steps and parameters are the same as those in Embodiment 1.

specific Embodiment approach 3

[0032] Specific embodiment three: the difference between this embodiment and specific embodiment one or two is that in the step five, the position of the simulated target is selected as the area of ​​the target to be added in the Gaofen No. 4 multispectral image of the simulated target to be added, This area should meet the objective conditions for the existence of the target (for example, the ship can no longer be on land), and the Poisson fusion algorithm is used to add the multi-spectral target map of Gaofen-2 processed in step 4 to the simulated target selected in the multi-spectral image of Gaofen-4. position, to obtain the multi-spectral image of Gaofen-4 with the added simulation target; the specific process is as follows:

[0033] Poisson fusion is an image stitching method based on the image gradient field. The GF-2 multi-spectral target image processed in step 4 and the GF-4 multi-spectral image to be added to the simulation target are divided into five according to t...

Images