CN106441808A - Thermal infrared hyperspectral imager blind pixel detection device and method - Google Patents

Abstract

The invention relates to a thermal infrared hyperspectral imager blind pixel detection device and method. The device comprises a blackbody controller, a standard blackbody, a thermal infrared hyperspectral imager and a computer module. The method includes the steps that firstly, the thermal infrared hyperspectral imager aims at the standard blackbody and fills a view field, the blackbody is set to be at N different temperatures through the blackbody controller, radiation light emitted after the temperatures are stabilized is received by the thermal infrared hyperspectral imager, and the computer module controls the thermal infrared hyperspectral imager to collect N sets of data; secondly, the N sets of data are subjected to wave band superposition according to temperature rise to generate temperature rise blackbody thermal infrared hyperspectral data; thirdly, a non-blind-pixel temperature rise spectrum is collected, blind pixel detection is conducted in a spectral mapping way, a blind pixel is marked, and a blind pixel detection result is generated. Blind pixel high-precision detection of the thermal infrared hyperspectral imager is achieved from the spectral dimension perspective, and the method plays an important role in calibration and data preprocessing of the thermal infrared hyperspectral imager.

Description

A kind of thermal infrared hyperspectral imager blind element detection means and method

Technical field

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

Background technology

Due to infrared acquisition components and parts casting technique or environmental change reason, thermal infrared high spectrum image meeting generally existing is blind Unit, the thermal infrared hyperspectral imager picture quality that the presence of blind element is swept to linear array push causes to have a strong impact on, to radiation calibration essence Degree will also result in impact.Therefore before image radiation correction, blind element detection is necessary links, and the detection quality of blind element directly affects The data processing of pictures subsequent and image quality evaluation.

The method of blind element detection at present includes the method based on Laboratory Calibration method with based on scene, and these detection methods are many Carry out detection process, the mostly detection based on image space dimension, thermal infrared EO-1 hyperion for single-range face battle array infrared image Imager is the forward position load of current high light spectrum image-forming research, realizes hundreds of wave band imaging in same infrared focus plane, blind The presence of unit causes to have a strong impact on to this imaging mode picture quality, and the detection data based on whole figure processes and might not fit With.Therefore, it is also desirable to research is directed to specific device and the algorithm of thermal infrared hyperspectral imager blind element detection.

Content of the invention

Problem to be solved by this invention is：A kind of device being applied to the detection of thermal infrared hyperspectral imager blind element is provided And method, the method is detected from spectrum dimension angle, solves a difficult problem for current thermal infrared hyperspectral imager blind element detection.

A kind of be applied to thermal infrared hyperspectral imager blind element detection device by black matrix controller, standard blackbody, heat red Outer hyperspectral imager and computer module composition, black matrix controller is connected with standard blackbody, and computer module is high with thermal infrared Optical spectrum imagers connect, thermal infrared hyperspectral imager visual field alignment criteria black matrix；Black matrix controller established standardses blackbody temperature, Standard blackbody sends thermal infrared radiation, and thermal infrared hyperspectral imager receives thermal infrared radiation, and computer module controls imager Carry out data acquisition, obtain the thermal infrared radiation data of 8-12.5 μm of each wave band of thermal infrared, change the temperature of standard blackbody, such as This is reciprocal, obtains 8-12.5 μm of thermal infrared radiation data under different temperatures, realizes thermal infrared high light spectrum image-forming based on this device The blind element detection of instrument data,.

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

1：Thermal infrared hyperspectral imager is directed at black matrix and is full of visual field, by black matrix controller, blackbody temperature is arranged For N number of (N >=30) different temperature, after temperature stabilization, computer module controls imager collection N group data；

1.1 connect blind element detection means, thermal infrared hyperspectral imager is directed at black matrix and is full of visual field, thermal infrared is high Optical spectrum imagers probe unit size is X × Y, and space dimension pixel number is X, and spectrum ties up image band after being imaged for Y Number is Y；

1.2 arrange black matrix initial temperature for T by black matrix controller₁, after temperature stabilization, computer module controls heat red Outer hyperspectral imager collection one group of data of storage, data acquisition line number is more than 300 row；

1.3 pass through the setting of black matrix controller adjusts temperature, and the size adjusting temperature is Δ T, and temperature stabilization is treated in Δ T≤5 DEG C Computer module controls one group of new data of thermal infrared hyperspectral imager collection storage afterwards, so repeats n times (N >=30), becomes As spectrogrph obtains the black body radiation data at a temperature of N difference.

2：The N group data that collection is obtained carries out band overlapping according to temperature rise, generates temperature rise black matrix high-spectral data；

2.1 every group of data obtaining collection are averaging processing, average after image size be X × Y, each pixel Value D is：

In formula, D_iFor the value of averagely rear i-th pixel, m is the line number of image acquisition, d_iFor i-th pixel value of jth row.

The different temperatures image that N number of size is X × Y is obtained by average treatment：T₁

B(T)_X×Y, T=T₁,…,…T_N(T₁<T₂<<T_N)

The 2.2 N group data obtaining collection carry out band overlapping according to temperature rise, generate temperature rise black matrix high-spectral data, Fig. 3 Left figure is temperature rise black matrix high-spectral data schematic diagram, and the space dimension size of data is X, and spectrum dimension size is Y, and temperature is tieed up as N, carries Take the temperature rise spectrum of one pixel of left figure stain, such as right figure, transverse axis represents temperature, and the longitudinal axis represents the DN value of pixel；

3：Gather normal pixel temperature rise spectrum, blind element detection is carried out by Spectral matching approach, labelling blind element generates blind element Testing result.

Normal pixel temperature rise spectrum in 3.1 collection temperature rise black matrix thermal infrared high-spectral datas, asks flat to the spectrum being gathered All, the temperature rise spectrum after obtaining averagely, is designated as with reference to spectrum r=(r₁..., r_N)

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

In formula, t is temperature rise certain pixel curve of spectrum of black matrix thermal infrared high-spectral data, sets spectral modeling empirical value and divides Not Wei λ, carry out blind element differentiation：

All for normal for testing result pixel wave bands are averaging by 3.3 calculating, calculate the often mean μ of row image and standard deviation sigma, To every row, all pixels judge

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

Cumulative for the blind element of 3.2 and 3.3 detections merging is marked by 3.4, generates blind element testing result, completes blind element inspection Survey.

By above method, the blind element high accuracy that the present invention can realize thermal infrared hyperspectral imager probe unit is examined Survey, the detection method tieing up angle from spectrum can be prevented effectively from conventional method missing inspection and the deficiency of empty inspection, and this method is significantly Degree improves blind element accuracy of detection.

Brief description

Fig. 1 be thermal infrared hyperspectral imager blind element detection schematic device, in figure device include black matrix controller 1, Standard blackbody 2, thermal infrared hyperspectral imager 3 and computer module 4..

Fig. 2 is the operating process of thermal infrared hyperspectral imager blind element detection.

Fig. 3 is temperature rise thermal infrared high-spectral data schematic diagram.

Fig. 4 is thermal infrared EO-1 hyperion temperature rise curve schematic diagram.

Specific embodiment

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

Fig. 1 describes a kind of composition structure chart of the device being applied to the detection of thermal infrared hyperspectral imager blind element, this dress Put and be made up of black matrix controller 1, standard blackbody 2, thermal infrared hyperspectral imager 3 and computer module 4, connecting as illustrated should Device, black matrix controller 1 and standard blackbody 2 connect, and thermal infrared hyperspectral imager 3 and computer module 4 connect, standard blackbody 2 and thermal infrared hyperspectral imager 3 be aligned, wherein：

(1) temperature control to standard blackbody realized by black matrix controller 1, adjusts black matrix to different temperatures；

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

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

(4) computer module 4 comprises to realize data acquisition and blind element detection.

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 detection means according to Fig. 1 mode, by thermal infrared hyperspectral imager be aligned black matrix and be full of regard , thermal infrared hyperspectral imager probe unit size is 320 × 256, and space dimension pixel number is 320, and spectrum is tieed up as 256 After individual i.e. 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, computer module controls heat Infrared high spectrum imaging instrument collection one group of data of storage, data acquisition line number is 300 row；

1.3 pass through the setting of black matrix controller adjusts temperature, raises temperature level and is 2 DEG C, computer mould after temperature stabilization Block controls one group of new data of thermal infrared hyperspectral imager collection storage, so repeats 49 times, imaging spectrometer obtains 50 altogether Black body radiation data at a temperature of individual difference, 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 every group of data obtaining collection are averaging processing, average after image size be 320 × 256, each picture Unit value D be：

In formula, D_iFor the value of averagely rear i-th pixel, d_iFor i-th pixel value of jth row.

The different temperatures image that 50 sizes are 320 × 256 is obtained by average treatment.

2.2 50 groups of data obtaining collection carry out band overlapping according to temperature rise, generate temperature rise black matrix thermal infrared EO-1 hyperion Data, Fig. 3 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, Temperature is tieed up as 50, extracts the temperature rise spectrum of one pixel of left figure stain, such as Fig. 4, and transverse axis represents temperature, and the longitudinal axis represents the DN of pixel Value；

3：Carry out blind element detection using based on Spectral matching strategy

Normal pixel temperature rise spectrum in 3.1 collection temperature rise black matrix high-spectral datas, is averaging to the spectrum being gathered, obtains Temperature rise spectrum after average, is designated as with reference to spectrum r=(r₁..., r_N).

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

In formula, t be certain pixel curve of spectrum of temperature rise black matrix high-spectral data, set spectral modeling empirical value as λ= 0.05, carry out blind element differentiation：

All for normal for testing result pixel wave bands are averaging by 3.3 calculating, calculate the often mean μ of row image and standard deviation sigma, To every row, all pixels judge

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

Cumulative for the blind element of 3.2 and 3.3 detections merging is marked by 3.4, generates blind element testing result, completes blind element inspection Survey.

Relative radiometric calibration coefficient is calculated according to testing result, Nonuniformity Correction is carried out to image.

Claims (2)

1. a kind of thermal infrared hyperspectral imager blind element detection means, device includes black matrix controller (1), standard blackbody (2), heat Infrared high spectrum imaging instrument (3) and computer module (4) it is characterised in that black matrix controller (1) is connected with standard blackbody (2), Computer module (4) is connected with thermal infrared hyperspectral imager (3), thermal infrared hyperspectral imager visual field alignment criteria black matrix； Black matrix controller (1) established standardses blackbody temperature, standard blackbody (2) sends thermal infrared radiation, thermal infrared hyperspectral imager (3) Receive thermal infrared radiation, computer module (4) controls imager to carry out data acquisition, obtains 8-12.5 μm of each wave band of thermal infrared Thermal infrared radiation data, change standard blackbody (2) temperature, and so on, obtain different temperatures under 8-12.5 μm of heat red External radiation data, ties up, from spectrum, the collection that angle realizes thermal infrared hyperspectral imager blind element detection data based on this device.

2. a kind of thermal infrared hyperspectral imager based on thermal infrared hyperspectral imager blind element detection means described in claim 1 The method of blind element detection is it is characterised in that comprise the following steps：

(1) thermal infrared hyperspectral imager is directed at black matrix and is full of visual field, blackbody temperature is set to by N by black matrix controller Individual, N >=30, different temperature, after temperature stabilization, computer module controls imager collection N group data；

(2) the N group data that collection obtains is carried out band overlapping according to temperature rise, generate temperature rise black matrix thermal infrared high-spectral data； Data genaration comprises the following steps that：

(2-1) every group of data obtaining collection is averaging processing, average after image size be X × Y, the value of each pixel D is：

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

In formula, D_iFor the value of averagely rear i-th pixel, m is the line number of image acquisition, d_iFor i-th pixel value of jth row；

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

(2-2) the N group data that collection obtains is carried out band overlapping according to temperature rise, generate temperature rise black matrix thermal infrared EO-1 hyperion number According to；

(3) gather normal pixel temperature rise spectrum, blind element detection is carried out by Spectral matching approach, labelling blind element generates blind element detection Result；Comprise the following steps that：

(3-1) normal pixel temperature rise spectrum in collection temperature rise black matrix thermal infrared high-spectral data, is averaging to the spectrum being gathered, Temperature rise spectrum after obtaining averagely, is designated as with reference to spectrum r=(r₁,…,r_N)；

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

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

In formula, t is temperature rise certain pixel curve of spectrum of black matrix thermal infrared high-spectral data, sets spectral modeling empirical value as λ, enters Row blind element differentiates：

(3-3) all for normal for testing result pixel wave bands are averaging, calculate the often mean μ of row image and standard deviation sigma, by being about to All pixels are judged

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

(3-4) the cumulative merging of blind element that step (3-2) and step (3-3) detect is marked, generates blind element testing result, complete Become blind element detection.