CN105372040A - Detection device and detection method of blind pixels of thermal infrared hyperspectral imager - Google Patents

Abstract

The present invention relates to a detection device and detection method of blind pixels of a thermal infrared hyperspectral imager. The detection device comprises a blackbody controller, a radiation blackbody, a thermal infrared hyperspectral imager and a computer module. The computer module includes a data acquisition unit and a blind pixel detection unit. Through adoption of the detection device provided by the invention, the detection method comprises the following steps: (1) allowing the thermal infrared hyperspectral imager to aim at the blackbody and be full of a field, the temperature of the blackbody is set to N (N is not less than 30) different temperatures through the blackbody controller, and the data acquisition unit is configured to control the imager to acquire N groups of data after the temperature is stabilized; (2) N groups of data is subjected to wave band superposition according to temperature rising, and temperature rising blackbody thermal infrared hyperspectral data is generated; and (3) a non-blind pixel temperature rising spectrum is acquired, the detection of blind pixels is performed through the way of spectrum matching, and the blind pixels are marked to generate the detection result of the blind pixels. According to the invention, from the spectral dimensionality point of view, the high precision detection of blind pixels of a thermal infrared hyperspectral imager may be realized so as to have an important effect on the calibration and data preprocessing of the thermal infrared hyperspectral imager.

Description

A kind of thermal infrared hyperspectral imager blind element pick-up unit and method

Technical field

The invention belongs to remote sensing and imaging spectrometer calibrates field, particularly the apparatus and method that detect of a kind of thermal infrared hyperspectral imager blind element.

Background technology

Due to infrared acquisition components and parts casting technique or environmental change reason, thermal infrared high spectrum image meeting ubiquity blind element, the existence of blind element causes the thermal infrared hyperspectral imager picture quality that linear array push is swept and has a strong impact on, and also can impact radiation calibration precision.Therefore before image radiation corrects, blind element detects is necessary links, and the detection quality of blind element directly affects data processing and the image quality evaluation of pictures subsequent.

The method that current blind element detects comprises based on Laboratory Calibration method and the method based on scene, these detection method spininess carry out check processing to single-range battle array infrared image, mostly be the detection based on image space dimension, thermal infrared hyperspectral imager is the forward position load of current high light spectrum image-forming research, a hundreds of wave band imaging is realized at same infrared focus plane, the existence of blind element causes this imaging mode picture quality and has a strong impact on, and might not be suitable for based on the detection of whole figure and data processing.Therefore, also need to study the specific device for the detection of thermal infrared spectrum imager blind element and algorithm.

Summary of the invention

Problem to be solved by this invention is: provide a kind of apparatus and method being applicable to thermal infrared hyperspectral imager blind element and detecting, and the method detects from spectrum dimension angle, solves the difficult problem that current thermal infrared hyperspectral imager blind element detects.

A kind of device being applicable to the detection of thermal infrared hyperspectral imager blind element is made up of black matrix controller, radiation black matrix, thermal infrared spectrum imager and computer module, wherein computer module comprises data acquisition unit and blind element detecting unit, data acquisition unit controls thermal infrared high light spectrum image-forming and carries out data acquisition, realize data record, and by different temperatures data genaration temperature rise black matrix thermal infrared high-spectral data; Blind element detecting unit detects based on spectrum dimension temperature rise black matrix thermal infrared high-spectral data, generates testing result.

Based on the method being applicable to the detection of thermal infrared hyperspectral imager blind element of this device, comprise the following steps:

1: thermal infrared hyperspectral imager aimed at black matrix and is full of visual field, by black matrix controller, blackbody temperature being set to N number of (N >=30) different temperature, after temperature stabilization, data acquisition unit controls imager and gathers N group data;

1.1 connect blind element pick-up unit, and thermal infrared hyperspectral imager is aimed at black matrix and is full of visual field, thermal infrared hyperspectral imager probe unit size is X × Y, and space dimension pixel number is X, and spectrum dimension is that after the individual i.e. imaging of Y, image band number is Y;

1.2 to arrange black matrix initial temperature by black matrix controller be T ₁, after temperature stabilization, data acquisition unit controls the collection of thermal infrared hyperspectral imager and stores one group of data, and data acquisition line number is greater than 300 row;

1.3 arrange adjustment temperature by black matrix controller, the size regulating temperature is Δ T, Δ T≤5 DEG C, after temperature stabilization, data acquisition unit controls the collection of thermal infrared hyperspectral imager and stores one group of new data, repetition like this N time (N >=30), imaging spectrometer obtains the blackbody radiation data at the different temperature of N.

2: utilizing data acquisition unit to carry out band overlapping by gathering the N group data obtained according to temperature rise, generating temperature rise black matrix high-spectral data;

2.1 utilize data acquisition unit to be averaging processing gathering the often group data obtained, and the image size after average is X × Y, and the value D of each pixel is:

$D_i=\frac{\underset{j=1}{\overset{m}{\Σ}}{d}_i}{X\×Y}$

In formula, D _ifor the value of average rear i-th pixel, m is the line number of image acquisition, d _ifor jth row i-th pixel value.

The different temperatures image that N number of size is X × Y is obtained: T by average treatment ₁

B(T) _X×Y,T＝T ₁,……，T _N(T ₁<T ₂<<T _N)

2.2 utilize data acquisition unit to carry out band overlapping by gathering the N group data obtained according to temperature rise, generate temperature rise black matrix high-spectral data, the left figure of Fig. 3 is temperature rise black matrix high-spectral data schematic diagram, the space dimension size of data is X, and spectrum dimension size is Y, and temperature dimension is N, extract the temperature rise spectrum of a left figure stain pixel, as right figure, transverse axis representation temperature, the longitudinal axis represents the DN value of pixel;

3: gather normal pixel temperature rise spectrum, carry out blind element detection by Spectral matching approach, mark blind element generates blind element testing result.

3.1 gather normal pixel temperature rise spectrum in temperature rise black matrix thermal infrared high-spectral data, are averaging, obtain the temperature rise spectrum on average, be designated as with reference to spectrum r=(r gathered spectrum ₁..., r _n)

3.2 calculate the spectral modeling with reference to spectrum and all pixels of temperature rise black matrix thermal infrared high-spectral data, and computing formula is

$\mathrm{\&delta;}={\cos }^{-1}\frac{\underset{i=1}{\overset{N}{\&Sigma;}}{t}_i{r}_i}{\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{t}_i^2}*\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{r}_i^2}}$

In formula, t is certain pixel curve of spectrum of temperature rise black matrix thermal infrared high-spectral data, and setting spectral modeling empirical value is respectively λ, carries out blind element differentiation:

3.3 calculate all wave bands of normal for testing result pixel are averaging, calculate average μ and the standard deviation sigma of often row image, to often go all pixels judge

In formula, i represents the i-th row image, i-th wave band after corresponding imaging

3.4 are merged into row labels by cumulative for 3.2 and 3.3 blind elements detected, and generate blind element testing result, complete blind element and detect.

By above method, the present invention can realize the blind element high precision test of thermal infrared hyperspectral imager probe unit, effectively can avoid the deficiency of the undetected and empty inspection of conventional method from the detection method of spectrum dimension angle, this method has increased substantially blind element accuracy of detection.

Accompanying drawing explanation

Fig. 1 is the device schematic diagram that thermal infrared hyperspectral imager blind element detects.

Fig. 2 is the operating process that thermal infrared hyperspectral imager blind element detects.

Fig. 3 is the result figure after single temperature-averaging.

Fig. 4 is the temperature rise thermal infrared spectrum schematic diagram data detected for blind element that image acquisition units generates.

Fig. 5 is the non-blind element temperature rise curve of spectrum, and transverse axis representation temperature (scope 0-98 DEG C), the longitudinal axis represents DN value.

Fig. 6 is thermal infrared EO-1 hyperion blind element testing result (white point).

Fig. 7 is the nonuniformity correction design sketch after blind element is repaired.

Embodiment

Below in conjunction with accompanying drawing, the present invention is described further.

Fig. 1 describes a kind of composition structural drawing being applicable to the device that thermal infrared hyperspectral imager blind element detects, and this device is made up of black matrix controller, radiation black matrix, thermal infrared spectrum imager and computer module, wherein:

(1) black matrix controller realizes controlling the temperature of radiation black matrix, regulates black matrix to different temperatures;

(2) radiation black matrix is thermal-radiating standard item, can launch stable radiation signal, and radiation visual field fit in this device should be greater than imaging spectrometer visual field;

(3) thermal infrared hyperspectral imager is instrument and equipment to be detected;

(4) computer module comprises data acquisition unit and blind element detecting unit, and data acquisition unit controls thermal infrared high light spectrum image-forming and carries out data acquisition, realizes data record, and generates temperature rise thermal infrared high-spectral data by the calibration data of different temperatures; Blind element detecting unit detects based on spectrum dimension temperature rise thermal infrared high-spectral data, generates testing result.

Below in conjunction with Fig. 1-Fig. 4, thermal infrared hyperspectral imager blind pixel detection method is described in detail.

1: imager gathers the black matrix data of 50 groups of different temperatures

1.1 connect blind element pick-up unit according to Fig. 1 mode, thermal infrared hyperspectral imager is aimed at black matrix and is full of visual field, thermal infrared hyperspectral imager probe unit size is 320 × 256, and space dimension pixel number is 320, and spectrum dimension is 256 is that after imaging, image band number is 256;

1.2 arrange black matrix initial temperature by black matrix controller is-10 DEG C, and after temperature stabilization, data acquisition unit controls the collection of thermal infrared hyperspectral imager and stores one group of data, and data acquisition line number is 300 row;

1.3 arrange adjustment temperature by black matrix controller, raised temperature size is 2 DEG C, after temperature stabilization, data acquisition unit controls the collection of thermal infrared hyperspectral imager and stores one group of new data, repetition like this 49 times, imaging spectrometer obtains the blackbody radiation data at 50 different temperature altogether, temperature range is 0-98 DEG C, 2 DEG C of intervals.

2: temperature rise black matrix thermal infrared high-spectral data generates

2.1 utilize data acquisition unit to be averaging processing gathering the often group data obtained, and the image size after average is 320 × 256, and the value D of each pixel is:

$D_i=\frac{\underset{j=1}{\overset{m}{\&Sigma;}}{d}_i}{320\&times;256}$

In formula, D _ifor the value of average rear i-th pixel, d _ifor jth row i-th pixel value.

Obtain by average treatment the different temperatures image that 50 sizes are 320 × 256, the average result at 20 DEG C is as Fig. 3.

2.2 utilize data acquisition unit to carry out band overlapping by gathering the 50 groups of data obtained according to temperature rise, generate temperature rise black matrix thermal infrared high-spectral data, the left figure of Fig. 4 is temperature rise black matrix thermal infrared high-spectral data schematic diagram, the space dimension size of data is 320, and spectrum dimension size is 256, and temperature dimension is 50, extract the temperature rise spectrum of a left figure stain pixel, as right figure, transverse axis representation temperature, the longitudinal axis represents the DN value of pixel;

3: utilize blind element detecting unit to utilize and carry out blind element detection based on Spectral matching strategy

3.1 gather normal pixel temperature rise spectrum in temperature rise black matrix high-spectral data, are averaging, obtain the temperature rise spectrum on average, be designated as with reference to spectrum r=(r gathered spectrum ₁..., r _n), as Fig. 5.

3.2 calculate the spectral modeling with reference to spectrum and all pixels of temperature rise black matrix high-spectral data, and computing formula is

$\mathrm{\&delta;}={\cos }^{-1}\frac{\underset{i=1}{\overset{N}{\&Sigma;}}{t}_i{r}_i}{\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{t}_i^2}*\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{r}_i^2}}$

In formula, t is certain pixel curve of spectrum of temperature rise black matrix high-spectral data, and setting spectral modeling empirical value is λ=0.05, carries out blind element differentiation:

3.3 calculate all wave bands of normal for testing result pixel are averaging, calculate average μ and the standard deviation sigma of often row image, to often go all pixels judge

In formula, i represents the i-th row image, i-th wave band after corresponding imaging

3.4 are merged into row labels by cumulative for 3.2 and 3.3 blind elements detected, and generate blind element testing result, complete blind element and detect, testing result is as Fig. 6.

Calculate relative radiometric calibration coefficient according to testing result, carry out Nonuniformity Correction to image, calibration result as shown in Figure 7.

Claims (2)

1. a thermal infrared hyperspectral imager blind element pick-up unit, device comprises black matrix controller, radiation black matrix, thermal infrared hyperspectral imager and computer module, it is characterized in that, wherein computer module comprises data acquisition unit and blind element detecting unit, data acquisition unit controls thermal infrared high light spectrum image-forming and carries out data acquisition, realize data record, and by different temperatures data genaration temperature rise black matrix thermal infrared high-spectral data; Blind element detecting unit detects based on spectrum dimension temperature rise black matrix thermal infrared high-spectral data, generates testing result.

2., based on the method that the thermal infrared hyperspectral imager blind element of thermal infrared hyperspectral imager blind element pick-up unit described in claim 1 detects, it is characterized in that comprising the following steps:

(1) thermal infrared hyperspectral imager aimed at black matrix and be full of visual field, blackbody temperature being set to N number of, N >=30, different temperature by black matrix controller, after temperature stabilization, data acquisition unit controls imager and gathers N group data;

(2) utilizing data acquisition unit to carry out band overlapping by gathering the N group data obtained according to temperature rise, generating temperature rise black matrix thermal infrared high-spectral data; Data genaration concrete steps are as follows:

(2-1) utilize data acquisition unit to be averaging processing gathering the often group data obtained, the image size after average is X × Y, and the value D of each pixel is:

$D_i=\frac{\underset{j=1}{\overset{m}{\&Sigma;}}{d}_i}{X\&times;Y}$

In formula, D _ifor the value of average rear i-th pixel, m is the line number of image acquisition, d _ifor jth row i-th pixel value;

The different temperatures image that N number of size is X × Y is obtained by average treatment:

(2-2) utilizing data acquisition unit to carry out band overlapping by gathering the N group data obtained according to temperature rise, generating temperature rise black matrix thermal infrared high-spectral data;

(3) gather normal pixel temperature rise spectrum, carry out blind element detection by Spectral matching approach, mark blind element generates blind element testing result; Concrete steps are as follows:

(3-1) gather normal pixel temperature rise spectrum in temperature rise black matrix thermal infrared high-spectral data, gathered spectrum is averaging, obtain the temperature rise spectrum on average, be designated as with reference to spectrum r=(r ₁..., r _n);

(3-2) calculate the spectral modeling with reference to spectrum and all pixels of temperature rise black matrix thermal infrared high-spectral data, computing formula is:

$\mathrm{\&delta;}={\cos }^{-1}\frac{\underset{i=1}{\overset{N}{\&Sigma;}}{t}_i{r}_i}{\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{t}_i^2}*\sqrt{\underset{i=1}{\overset{N}{\&Sigma;}}{r}_i^2}}$

In formula, t is certain pixel curve of spectrum of temperature rise black matrix thermal infrared high-spectral data, and setting spectral modeling empirical value is λ, carries out blind element differentiation:

(3-3) all wave bands of normal for testing result pixel being averaging, calculating average μ and the standard deviation sigma of often row image, judging by being about to all pixels

In formula, i represents the i-th row image, i-th wave band after corresponding imaging

(3-4) blind element step (3-2) and step (3-3) detected is cumulative is merged into row labels, generates blind element testing result, completes blind element and detects.