CN102829867B - Simulation method of hyperspectral remote sensing data embedded based on target empty spectral information - Google Patents

Abstract

The invention discloses a simulation method of hyperspectral remote sensing data embedded based on target empty spectral information, which belongs to the technical field of simulation of remote sensing spectroscopic data, and solves the problem of shortage of a target in hyperspectral remote sensing data in the conventional remote sensing simulation technology. The simulation method comprises the following steps of: step 1, generating target space information; step 2, generating target spectral information; step 3, combining the information generated in the step 1 and the information generated in the step 2 to generate target empty spectral information; and step 4, embedding the target empty spectral information into hyperspectral background information so as to form the hyperspectral remote sensing data comprising the target. The simulation method is suitable for simulating the hyperspectral remote sensing data.

Description

The high-spectrum remote sensing data emulation mode that the empty spectrum information of based target embeds

Technical field

The present invention relates to the high-spectrum remote sensing data emulation mode that the empty spectrum information of a kind of based target embeds, belong to Remote Spectra data simulation technical field.

Background technology

Remote sensing refers generally to use the electromagnetic radiation of sensor to object, the detection of reflection characteristic, and the science and technology of the theory that the character to object, feature and state are analyzed according to its characteristic, methods and applications.Remote sensing resources generally refers to and adopts Aeronautics and Astronautics delivery vehicle, the different types of data obtaining by sensor, such as visible ray, multispectral, high spectrum, SAR and infrared etc., phase and different resolution when different, such as comprising the data of temporal resolution, spatial resolution and spectral resolution.Remote sensing resources has now been widely used in national economy every field, is with a wide range of applications.

Remote sensing technology develop rapidly, its system becomes increasingly complex, kind is more and more various, and these have brought many difficulties to system.High-spectrum remote-sensing is a kind of novel remote sensing technology, and it is the combination of spectral technique and imaging technique, is also the important new direction of development of remote sensing.Lack at present the high-spectrum remote sensing data that contains target, if can obtain the emulated data of adding after target, for determining significant of the design of system index and the range of application of system.

Summary of the invention

The present invention is the problem that lacks target in the high-spectrum remote sensing data in existing remote sensing emulation technology in order to solve, and a kind of based target high-spectrum remote sensing data emulation mode that empty spectrum information embeds is provided.

The high-spectrum remote sensing data emulation mode that the empty spectrum information of based target of the present invention embeds, it comprises the following steps:

Step 1: generate object space information;

Step 2: generate target optical spectrum information;

Step 3: the information of step 1 and step 2 generation is synthesized and generated target empty spectrum information;

Step 4: this target empty spectrum information is embedded in high spectral background information, forms the high-spectrum remote sensing data that comprises target.

The concrete grammar that generates object space information in described step 1 is:

Select target spatial form template in target shape template base, the spatial information of this template is that the mode by taking pictures generates; Then take the spatial resolution of high spectral background information as benchmark, object space shape template is carried out to spatial sampling, to reach the consistent of the two spatial resolution, thereby generate object space information.

The concrete grammar that generates target optical spectrum information in described step 2 is:

According to the material of target corresponding to the information of object space described in step 1, in library of spectra, choose the reflectance spectrum after corresponding normalization, generate target optical spectrum information.

The concrete grammar that in described step 3, the information of step 1 and step 2 generation is synthesized to generation target empty spectrum information is:

One by one normalized gray-scale value corresponding each pixel in spatial form template be multiplied by this pixel corresponding normalization after reflectance spectrum, form normalized target empty spectrum information, normalized target empty spectrum information is multiplied by the maximum gradation value in high spectral background information, generates target empty spectrum information.

In described step 4, this target empty spectrum information is embedded in high spectral background information, the concrete grammar that forms the high-spectrum remote sensing data that comprises target is:

Determine the to be embedded position of target empty spectrum information in high spectral background information, the respective pixel of this position to be embedded is replaced with to target empty spectrum information, and the edge of this target empty spectrum information is carried out to mean filter, form the high-spectrum remote sensing data that comprises target.

Advantage of the present invention is: the present invention is directed to the situation that lacks target in high-spectral data, the information that comprises object space characteristic and spectral characteristic is embedded in high spectral background data, the high spectrum emulated data that formation comprises target information, be conducive to the design of follow-up remote sensing system index and the range of application of system determine on meet application needs.

The inventive method has reached simulation result effectively and reasonably.In use change the spectral characteristic of scene according to the library of spectra of setting up, can emulation generate the spectral characteristic of target under various material, for the development of high-spectrum remote-sensing device provides reference.

Accompanying drawing explanation

Target empty spectrum information is embedded into the principle schematic in high spectral background information by Fig. 1;

Fig. 2 is the generation block diagram of the high-spectrum remote sensing data that comprises target.

Embodiment

Embodiment one: below in conjunction with Fig. 1 to Fig. 2, present embodiment is described, the high-spectrum remote sensing data emulation mode that the empty spectrum information of present embodiment based target embeds, is characterized in that: it comprises the following steps:

Step 1: generate object space information;

Step 2: generate target optical spectrum information;

Step 3: the information of step 1 and step 2 generation is synthesized and generated target empty spectrum information;

Step 4: this target empty spectrum information is embedded in high spectral background information, forms the high-spectrum remote sensing data that comprises target.

Embodiment two: present embodiment is to the further illustrating of embodiment one, the concrete grammar that generates object space information described in present embodiment in step 1 is:

Select target spatial form template in target shape template base, the spatial information of this template is that the mode by taking pictures generates; Then take the spatial resolution of high spectral background information as benchmark, object space shape template is carried out to spatial sampling, to reach the consistent of the two spatial resolution, thereby generate object space information.

The generation of object space information in present embodiment: both comprised spatial information in high spectral background information and also comprised spectral information, the target therefore embedding also will comprise this two category information.What first need generation is the spatial information of target, comprises again two parts: the selection of spatial form template and the generation of half-tone information in the generative process of object space information.

Can be on the basis of investigation in early stage, select to there is the typical target of the meaning of representing, and generate template base and the corresponding gray-scale value of each object pixel of this template of interesting target.Obtaining of half-tone information can, according to different target common reflected energy within the scope of panchromatic wave-band, can generate by the mode of taking pictures conventionally.Target in template base needs the transformation of a spatial resolution to the target in high-spectral data, in this process, can reasonably sample according to the spatial resolution of the object space resolution in template base and high spectral background data, to reach the consistent of spatial resolution.In telescopiny, need to from template base, choose the space template of target to be added and corresponding half-tone information is normalized.

Embodiment three: present embodiment is to the further illustrating of embodiment two, the concrete grammar that generates target optical spectrum information described in present embodiment in step 2 is:

According to the material of target corresponding to the information of object space described in step 1, in library of spectra, choose the reflectance spectrum after corresponding normalization, generate target optical spectrum information.

In present embodiment, determining on the basis of To Template and half-tone information, generate the corresponding spectral signature in target area.

First need to set up the library of spectra of target, the foundation of library of spectra can realize by the spectrum of actual measurement all types of target, also can realize by the mode of obtaining the ripe library of spectra that obtains widespread use.The generation of spectral information need to be chosen corresponding spectrum according to the material of target reality in library of spectra, it should be noted that the reflectance spectrum that will choose after normalization.

Embodiment four: present embodiment is described below in conjunction with Fig. 1 and Fig. 2, present embodiment is to the further illustrating of embodiment three, and the concrete grammar that the information in step 3, step 1 and step 2 being generated described in present embodiment synthesize generation target empty spectrum information is:

One by one normalized gray-scale value corresponding each pixel in spatial form template be multiplied by this pixel corresponding normalization after reflectance spectrum, form normalized target empty spectrum information, normalized target empty spectrum information is multiplied by the maximum gradation value in high spectral background information, generates target empty spectrum information.

In present embodiment, the information of first two steps is synthesized and is embedded in high spectral background data.

The generation of target empty spectrum information generates according to spectral information in locus, target place, and its implementation procedure is that normalized gray-scale value corresponding each pixel is multiplied by normalized reflectance spectrum, forms normalized target empty spectrum information.Then normalized empty spectrum information being multiplied by maximum gradation value in high spectral background data, if this maximum gradation value quantization digit is 12, is 4095,2 ¹²-1, generate target empty spectrum information.

Embodiment five: present embodiment is described below in conjunction with Fig. 1 and Fig. 2, be further illustrating embodiment four below in conjunction with present embodiment, described in present embodiment, in step 4, this target empty spectrum information is embedded in high spectral background information, the concrete grammar that forms the high-spectrum remote sensing data that comprises target is:

Determine the to be embedded position of target empty spectrum information in high spectral background information, the respective pixel of this position to be embedded is replaced with to target empty spectrum information, and the edge of this target empty spectrum information is carried out to mean filter, form the high-spectrum remote sensing data that comprises target.

In present embodiment, the imaging process of remotely-sensed data can be expressed as:

$I(x,y)={\∫}_{{\mathrm{\λ}}_0}^{{\mathrm{\λ}}_1}r(x,y,\mathrm{\λ})\×v\left(\mathrm{\λ}\right)\mathrm{d\λ},$

The reflection strength that wherein r (x, y, λ) is atural object to be imaged, v (λ) is the relative efficiency function under a certain wavelength.λ ₀and λ ₁for the wavelength that sensor can receive, wherein λ ₀for initial wavelength, λ ₁for cutoff wavelength.If the spatial data of target is imaging in visible-range, λ ₀be 0.4 μ m, λ ₁be 0.7 μ m; X is space transverse axis coordinate, and y is space ordinate of orthogonal axes, and λ is wavelength, the energy that I (x, y) receives for sensor.

And for high light spectrum image-forming, have up to a hundred components:

$I_i(x,y)={\∫}_{{\mathrm{\λ}}_i}^{{\mathrm{\λ}}_i+{\mathrm{\Δ\λ}}_i}r(x,y,\mathrm{\λ})\×v\left(\mathrm{\λ}\right)\mathrm{d\λ},i=\mathrm{1,2},...,N$

For scene simulation, needing to be processed is r (x, y, λ).Target embed process in, most critical be to handle background r well _b(x, y, λ) and target r to be embedded _trelation between (x, y, λ).For high-spectral data, not only need to consider spatial information, also need to consider spectral information; I _i(x, y) is i the energy that wave band sensor receives, λ _ibe the initial wavelength of i wave band sensor, Δ λ _iit is the waveband width of i wave band sensor.The spatial spectral information of target is embedded into the schematic diagram that goes in high spectral background data as shown in Figure 1.

In target empty spectrum information telescopiny, first need space and the spectral resolution of clear and definite background, take this as a foundation and process respectively spatial information and the spectral information of target to be embedded, complete the unification of the empty spectrum information of target, background, then target empty spectrum information is embedded in high spectral background information to the final high-spectral data that comprises target that forms.

For the target of high-spectral data embeds, be divided into three parts: the generation of object space information, the generation of target optical spectrum information and empty spectrum information are combined embedding, and telescopiny as shown in Figure 2.

Present embodiment has realized the embedding of combining of target empty spectrum information:

The position of selected target to be embedded in high spectral background information, the sky that is then target the high spectral background data replacement of respective pixel spectrum data.In order to reduce the inconsistency of target gray scale and background gray scale, the edge that embeds target is carried out to mean filter, make the effect of the more approaching actual data that obtain of generated data.

Through above four steps, just corresponding target can be joined in high spectral background data and goes, form the high-spectral data that comprises target.These data can be tested accordingly, determine the rationality of high-spectrum remote-sensing index of correlation, for development and the index demonstration of high-spectrum remote-sensing device provide reference.

Claims (4)

1. the high-spectrum remote sensing data emulation mode that the empty spectrum information of based target embeds, it comprises the following steps:

Step 1: generate object space information;

Step 2: generate target optical spectrum information;

Step 3: the information of step 1 and step 2 generation is synthesized and generated target empty spectrum information;

Step 4: this target empty spectrum information is embedded in high spectral background information, forms the high-spectrum remote sensing data that comprises target; It is characterized in that:

The concrete grammar that generates object space information in described step 1 is:

Select target spatial form template in target shape template base, the spatial information of this template is that the mode by taking pictures generates; Then take the spatial resolution of high spectral background information as benchmark, object space shape template is carried out to spatial sampling, to reach the consistent of the two spatial resolution, thereby generate object space information.

2. the high-spectrum remote sensing data emulation mode that the empty spectrum information of based target according to claim 1 embeds, is characterized in that: the concrete grammar that generates target optical spectrum information in described step 2 is:

According to the material of target corresponding to the information of object space described in step 1, in library of spectra, choose the reflectance spectrum after corresponding normalization, generate target optical spectrum information.

3. the high-spectrum remote sensing data emulation mode that the empty spectrum information of based target according to claim 2 embeds, is characterized in that: the concrete grammar that in described step 3, the information of step 1 and step 2 generation is synthesized to generation target empty spectrum information is:

One by one normalized gray-scale value corresponding each pixel in spatial form template be multiplied by this pixel corresponding normalization after reflectance spectrum, form normalized target empty spectrum information, normalized target empty spectrum information is multiplied by the maximum gradation value in high spectral background information, generates target empty spectrum information.

4. the high-spectrum remote sensing data emulation mode that the empty spectrum information of based target according to claim 3 embeds, it is characterized in that: in described step 4, this target empty spectrum information is embedded in high spectral background information, the concrete grammar that forms the high-spectrum remote sensing data that comprises target is:

Determine the to be embedded position of target empty spectrum information in high spectral background information, the respective pixel of this position to be embedded is replaced with to target empty spectrum information, and the edge of this target empty spectrum information is carried out to mean filter, form the high-spectrum remote sensing data that comprises target.

Images