RU2297728C1 - Method for correcting dissimilarity of multi-element photo-receiving devices with scanning - Google Patents

Abstract

FIELD: engineering of optical-electronic systems for forming and processing infrared images, may be used in thermal vision systems with scanning multi-row matrix photo-receiving devices.

SUBSTANCE: in accordance to method, successive registration of scene elements is performed by adjacent photo-sensitive areas during scanning, to perform correction. In the range of alternation of input signals from different parts of scene, dependence of signals of each element on signals of adjacent element is determined during registration of similar elements of scene (pixels). On basis of these ratios, correcting function is determined and correction of signals of each following element of multi-row matrix photo-receiving device is performed relatively to previous one in such a way that corrected signals of all elements of multi-row matrix photo-receiving devices are similar when color signals are similar in whole range of signals of scene.

EFFECT: possible correction of dissimilarity of multi-row matrix photo-receiving devices with scanning without usage of standard signal sources.

2 cl, 2 dwg

Description

The invention relates to optical-electronic systems for the formation and processing of infrared images, for which the urgent task of eliminating the heterogeneity of the signals due to differences in sensitivity to the input stream and in the dark current of the elements of photodetector devices (FPU). The invention can be used in thermal imaging systems with scanning when using multi-row FPU.

The known method [1, 2] for correcting the heterogeneity of multi-element MFPs in scanning thermal imagers (we consider it as an analogue), which consists in scanning the scene, recording the signals of the FPU elements from the observed scene, using the signal samples to determine the average correction coefficients by averaging, and the correction is carried out by multiplying the values of the samples of the signals by the corresponding line correction factors.

The disadvantage of this method is that only one correction factor is used to compensate for the signal, the uniformity of which depends on two parameters - the sensitivity of the element and on the bias due, in particular, to the dark current. In the case of simple scenes, with a smooth change in brightness, averaging window sliding over the image is sufficient to maintain the corrective ability. As the window moves, the corrective row coefficients change values characteristic of high brightness to corresponding low brightness. As scenes become more complex, the compensation of heterogeneity is hampered by the need to use more complex averaging windows. This factor limits the applicability of the method.

The closest in technical essence to the claimed invention is a method [3] of two-point correction, which provides compensation for sensitivity and displacement regardless of the scene, as it uses the signals of two reference sources, high and low radiation intensity, to determine correction coefficients. The method consists in scanning the scene and registering samples of the signals of the elements of the photodetector device, scanning reference sources and registering the samples of their signals, determining the corrective actions from the samples of the signals of the reference sources - coefficients for correction according to the sensitivity and offset of the elements of the photodetector, correcting the samples of the signals of the scene - subtract bias correction coefficients from sample values and multiply differences by sensitivity coefficients .

The essence of the method is as follows. When exposed to a reference optical signal of a low level, the same for all elements of the FPU, the average values of the electrical signals of low level for all elements of the MPU are

- среднее значение сигнала низкого уровня от i-го элемента,Where

- the average value of the low level signal from the i-th element,

i = 1,2, ..., N, N is the number of elements of the MFP;

T is the averaging interval, t are the signal sample numbers;

- сигнал i-го элемента на t-м отсчете;

- signal of the i-th element at the t-th sample;

And _i and B _i - sensitivity and displacement of the i-th element;

P ⁿ - low-level reference radiation flux;

E _i is the element noise averaged over the interval T, T is chosen large enough so that the influence of this component can be neglected.

When exposed to a high-level reference optical signal that is the same for all elements of the MFP, the average values of high-level electrical signals are

where P ⁱⁿ - a stream of reference radiation of a high level.

When exposed to arbitrary signals of the scene, the current signals of the individual elements S _i (t) are equal

, определяемые следующим образом:The correction of the scene signals is carried out by replacing the original scene signals S _i (t) with the corrected scene signals

defined as follows:

where ΔP ^int = (P ⁱⁿ -P ⁿ ) is the difference between the reference flows of high and low levels, i = 1,2, ..., N.

- коэффициенты по смещению, и

- по чувствительности. Применение этих коэффициентов к сигналам сцены - вычитание из сигналов коэффициентов по смещению и умножение разностей на коэффициенты по чувствительности - обеспечивает однородность скорректированных сигналов: одну и ту же по всем элементам чувствительность - величина 1/ΔP^вн, и одно и то же смещение, равное (-P^н/ΔP^вн).The coefficients correcting heterogeneity are

are the bias factors, and

- by sensitivity. The application of these coefficients to the signals of the scene - subtracting the coefficients of the offset from the signals and multiplying the differences by the sensitivity coefficients - ensures the uniformity of the corrected signals: the same sensitivity across all elements is 1 / ΔP ^int , and the same offset is equal to ( -P ⁿ / ΔP ^int ).

The main disadvantage is that the use of sources of reference signals is associated with the need for their adjustment to the scene. The method provides correction provided that the signals of the elements are linearly dependent on the optical flux at their input. In order to minimize deviations from the linear dependence within the dynamic range of the observed scene, the range of values of the reference flows is consistent with the range of values of the signals of the scene. In addition, it is required to maintain uniformity of illumination of MFPU elements by reference sources, since the accuracy of determination of correction factors depends on the uniformity of illumination. In some cases, the presence of reference sources is unacceptable under the terms of their placement in the equipment.

The aim of the present invention is to obtain the possibility of correcting the heterogeneity of multi-element photodetectors with scanning according to linear, two-point and non-linear correction schemes without using sources of reference signals.

The goal is achieved due to the fact that they scan the scene and register the samples of the signals of the elements of the photodetector, determine the corrective actions from the samples of the observed scene, and correct the samples of the signals of the scene. The scene is scanned in a sequence that ensures that the streams of the same scene elements hit neighboring elements of the photodetector and the streams of neighboring scene elements hits the same device element, for each pair of neighboring elements of the photodetector, i-th and i + 1-st , by the set of pairs of samples of signals S _i (t) and S _{i + 1} (t), i∈N, N is the set of elements of the photodetector and t is the time parameter, where each pair of samples corresponds to the same state of the scene element, s using approximation I determine corrective actions - the function of communication signals of neighboring elements

, где S_i(t) - отсчеты сигналов i-го элемента,

- скорректированные сигналы i-го элемента относительно j-го элемента, i, j∈N,so that these functions have the smallest standard deviation on the sets of pairs of samples S _i (t) and S _{i + 1} (t) and map any value from the ranges of variables s _i corresponding to the range of the signal S _i (t) into the values of the variables s _{i +1} , corresponding to the values from the signal ranges S _{i + 1} (t), correct the samples of the signals of the scene S _i (t) by replacing them with

where S _i (t) are the samples of the signals of the i-th element,

- the corrected signals of the i-th element relative to the j-th element, i, j∈N,

.where F _{j, i} (S ((t)) is the corrective action is the signal correction function S _i (t) relative to the j-th element, which is obtained by the sequence of application - the initial argument S _i (t), the communication functions F _{i + 1, i} (), F _{i + 2, i + 1} (), ..., F _{j, j-1} () signals of pairs of neighboring elements i + 1, i; i + 2, i + 1; ... j, j-1, making up the path from element i to element j, for i = j

.

The essence of the method is as follows.

The condition for scanning the scene — the flows of the same scene elements to neighboring elements of the photodetector — allows one to select pairs of samples from the same scene elements from the sequences of samples S _i (t) and S _{i + 1} (t) with the same state ( identical flows) and from the set of such pairs to determine, using approximation, corrective actions - the communication functions of neighboring elements F _{i + 1} , _i (),

которые отображают значения отсчетов i-го элемента на значения отсчетов элемента i+1. Отображение означает, что значению s_i отсчета сигнала S_i(t), которому соответствует некоторое значение потока Р на его входе, ставится в соответствие значение s_i+1 отсчета сигнала S_i+1(t), которому соответствует то же самое значение потока Р на его входе.

which map the values of the samples of the i-th element to the values of the samples of the element i + 1. The display means that the value s _{i of the} reference signal S _i (t), which corresponds to a certain value of the stream P at its input, is associated with the value s _{i + 1 of the} reference signal S _{i + 1} (t), which corresponds to the same value of the stream P at its entrance.

The scanning condition — the fluxes of neighboring scene elements to the same element of the photodetector — ensures the completeness of the set of communication functions of elements F _{i + 1} , _i () such that for all paths that connect all other elements of the device to any element of the photodetector, the communication functions of all pairs of neighboring elements of these paths are determined.

The procedure for adjusting the elements of the photodetector relative to the jth element is to determine the values of the correction functions F _{j, i} (S _i (t)) of the signals S _i (t), i, j∈N, each is determined by the sequence of applications of the communication functions (sequence corrective actions) on the way from the i-th to the j-th element.

- скорректированное значение сигнала S_i(t) относительно элемента i+1,The first in the sequence of applications of communication functions - application of the function F _{i + 1, i} (S _i (t)) - gives the value

- the adjusted value of the signal S _i (t) relative to the element i + 1,

Since the communication function is determined on the sequence of scene elements common to the i-th and i + 1-th elements of the MFP, substituting the argument S _i (t) into the function, which corresponds to a certain value of the scene element flow, gives the value S as the result of the function _{i + 1} (t), which corresponds to the same flow value.

Similarly, for subsequent applications, 1 = 1,2, ..., j-1,

- скорректированные относительно элементов i+1 значения сигнала S_i(t), и которые, следуя определениям функций связи, равны S_i(t). Начальному аргументу применения функций - S_i(t), которому соответствует некоторое значение потока элемента сцены, функции связи ставят в соответствие значения S_i+l(t), которым соответствует то же самое значение потока, что и сигналу S_i(t).which define

- signal values S _i (t) adjusted relative to elements i + 1, and which, following the definitions of communication functions, are equal to S _i (t). The initial argument of the application of the functions is S _i (t), which corresponds to a certain value of the flow of the scene element, the communication functions associate the values of S _{i + l} (t), which correspond to the same value of the stream as the signal S _i (t).

This means that the signals of the remaining elements corrected with respect to one of the elements of the photodetector will be the same for the same scene signals for all elements of the MFP, which corresponds to the purpose of the correction.

In the case of small changes in the signals within the scene, the dependences S _{i + 1} (t) on S _i (t) are close to linear and are approximated by direct

having the smallest standard deviation

on the measured dependences S _{i + 1} (t) on S _i (t).

Corrective actions: linear dependencies coefficients A _{i + 1, i} - the ratio of the sensitivities of neighboring elements, and B _{i + 1, i} - the relative displacement of neighboring elements, are determined by the formulas

where T _{i + 1, i} , is the set of pairs of samples S _i (t) and S _{i + 1} (t) according to the same scene elements, ordered by time: the value t = 1 corresponds to the first time sample, t = 2 - the next and so on to T _{i + 1, i} -th reference;

S _i and S _{i + 1} are the average values of the signals of the i-th and i + 1-th elements on the interval T _{i + 1, i} . With increasing interval, the ratio A _{i + 1, i} tends to A _{i + 1} / A _i .

Formulas for determining the relations of sensitivities and relative displacements of pairs of neighboring elements are obtained from linear regression of samples of signals S _i (t) from samples S _{i + 1} (t) and linear regression of samples S _{i + 1} (t) from S _i (t). To eliminate the bias of the values of A _{i + 1, i} with respect to A _{i + 1} / A _i due to the accumulation of noise in the formulas, we use the multiplication of neighboring, uncorrelated in noise, samples.

элементов i=1,2,...,N определяются по исходным, некорректированным сигналам S_i(t) и коэффициентам A_i+l,i+l-1 и В_i+l,i+l-1, 1=1,2,...,j-i, следующим образом:Signals corrected for element j

elements i = 1,2, ..., N are determined by the initial, uncorrected signals S _i (t) and the coefficients A _{i + l, i + l-1} and B _{i + l, i + l-1} , 1 = 1 , 2, ..., ji, as follows:

Figure 1 shows, in comparison with the implementation of the prototype method [3], a diagram of a variant of the device that implements the proposed method,

Where

1 - reference signal of a low level;

2 - high level signal standard;

3 - scanner;

4 - reference signal of a low level;

5 - scene signal;

6 - reference signal of a high level;

7 - photodetector;

8 - analog-to-digital signal converter;

9 - determinant of correction factors;

10 - scene signal corrector.

элементов i=1,2,...,N по исходным, некорректированным сигналам S_i(t) и коэффициентам A_i+l,i+l-1 и B_i+l,i+l-1, l=1,2,...,j-i и заменяет S_i(t) на

. В прототипе в корректоре вычитают из значений отсчетов сцены корректирующие коэффициенты по смещению и умножают разности на корректирующий коэффициент по чувствительности и этим скорректированным значением заменяют исходные значения отсчетов сцены.The dashed line marks the devices and signals necessary when using the prototype method and not needed for the proposed method. In the proposed method, the correction coefficients A _{i + 1, i} and B _{i + 1, i are} determined by the signals of the scene determinant of correction factors. In the prototype, this block determines the correction factors for sensitivity and bias using reference sources. In the proposed method, the scene signal corrector determines the signals corrected for element j

elements i = 1,2, ..., N according to the initial, uncorrected signals S _i (t) and the coefficients A _{i + l, i + l-1} and B _{i + l, i + l-1} , l = 1, 2, ..., ji and replaces S _i (t) by

. In the prototype, in the corrector, correction coefficients by offset are subtracted from the values of the scene samples and the differences are multiplied by a correction coefficient by sensitivity, and the original values of the scene samples are replaced with this adjusted value.

The condition for scanning the scene in a sequence that ensures that flows of the same scene elements to adjacent elements of the photodetector and hit flows of neighboring scene elements to the same element of the device, we will explain using figure 2, which shows the location of the photosensitive elements for multi-row 4 × N format. The image is scanned in the direction perpendicular to the lines of photosensitive elements A, B, ... F, each pixel of the scene image is recorded sequentially by elements A _i , B _i , C _i , D _i , and the neighboring pixel is recorded by elements E _i , F _i , G _i , H _i . The signals of the elements A _i , B _i , C _i , D _i can be adjusted with respect to the signals of any element of this group, and the signals of the elements E _i , F _i , G _i , H _i can be adjusted with respect to any element from of this group. At the second scan, the image can be shifted by half the step of the elements in the rulers. In this case, the sequence of pixels registered in the first scan by the elements A _i , B _i , C _i , D _i will be registered by the elements E _i , F _i , G _i , H _i . Thus, it is possible to correct the signals of the elements E _i , F _i , G _i , H _i with respect to the elements A _i , B _i , C _i , D _i , as well as the signals of the elements A _{i + 1} , B _{i + 1} , C _{i +1} , D _{i + 1} with respect to the elements E _i , F _i , G _i , N _i and thus all the elements of the MFP are connected by correction relations.

Information sources

1. RF patent 2113065. A method for equalizing the uneven sensitivity of photodetectors of scanning lines of thermal imagers, Belokonev VM, Degtyarev EV, Rudy IV, Malyshev IA, Pavlova VA, Teterin VV , Demesh O.V., Kabanov V.F. - Publ. 06/10/98, Bulletin No. 16.

2. Pavlova V.A. et al. "An Iconic Approach to Solving the Problem of Correcting the Sensitivity Inhomogeneities of Multi-element MFPs in Scanning Thermal Imagers", Optical Journal, Volume 64, No. 2, 1997.

3. David L. Perry, Eustase L. Dereniak. Linear theory of nonuniformity correction in infrared staring sensors. - "OPTICAL ENGINEERING", August 1993, Vol.32, No. 8, 1854-1859.

Claims (2)

1. The method of correcting the heterogeneity of multi-element photodetectors with scanning, namely, that they scan the scene and register the samples of the signals of the elements of the photodetector, correct the samples of the signals of the scene, characterized in that for the correction they are scanned in a sequence that ensures the flow of the same elements scenes on neighboring elements of the photodetector and the flow of neighboring elements of the scene on the same element of the device for each pair of neighboring ele ENTOV photodetector, i-th and (i + 1) -th, a plurality of pairs of signal samples S _i (t) and S _{i + 1} (t), i∈N, N - a plurality of photodetector elements and t - time parameter , and each pair of samples corresponds to the same state of the scene element, using approximation, corrective actions are determined - the communication functions of the signals of neighboring elements

so that these functions have the smallest standard deviation on the sets of pairs of samples S _i (t) and S _{i + 1} (t) and map any value from the ranges of variables s _i corresponding to the ranges of signals S _i (t) into the values of the variables s _{i +1} , corresponding to the values from the signal ranges S _{i + 1} (t), correct the samples of the signals S _i (t) by replacing them with

where S _i (t) are the samples of the signals of the i-th element,

corrected signals of the i-th element relative to the j-th element, i, j∈N,

where F _{j, i} (S _i (t)) is the signal correction function S _i (t) with respect to the jth element, which is obtained by sequential application of the arguments of the link functions F _{i + 1, i} (), F _{i + 2, i + 1} (), ..., F _{j, j-1} () signals of pairs of neighboring elements i + 1, i; i + 2, i + 1; ... j, j-1, making up the path from element i to element j, for i = j

= S _i (t), where S _i (t) is the argument of the first function, and the argument of each subsequent function is the result of the previous function, from F _{i + 1, i} () to F _{i, j-1} (). 2. The method according to claim 1, characterized in that the functions F _{i + 1, i} () are linear dependencies

having the smallest standard deviation

on the measured dependences S _{i + 1} (t) on S _i (t), corrective actions: the coefficients of linear dependences A _{i + 1, i} - the ratio of the sensitivities of neighboring elements and В _{i + 1, i} - the relative displacement of neighboring elements are determined by the formulas

where T _{i + 1, i} is the set of pairs of samples S _i (t) and S _{i + 1} (t) by the same scene elements, ordered by time: the value t = 1 corresponds to the first time sample, t = 2 - the next and so on to T _{i + 1, i-} th reference; S _i and S _{i + 1} are the average values of the signals of the i-th and (i + 1) -th elements in the interval T _{i + 1, i} , the signals corrected relative to the element j

elements i = 1, 2, ..., N are determined by the initial uncorrected signals S, (t) and the coefficients A _{i + l, i + l-1} and B _{i + l, i + l-1} , 1 = 1, 2, ..., ji, as follows:

Images