Image capture using radiation sensitive elements with memory effect.

Abstract

A method for capturing an image allows at least partially correcting the memory effect of the sensitive elements (1) of the matrix (10a) used to capture the image. The corrected image is formed by subtracting from a new captured raw image a portion of a previous raw image captured prior to the new image. The method is particularly suitable for sensitive elements with a linear transfer function in time, such as bolometers or microbolometers. Correcting the memory effect allows improving the transfer function and / or reducing tail effects that appear in the image when scene elements move.

Description

Detailed Description of the Invention

[0001] [Technical field] The present description relates to a method for capturing an image using a sensitive element (sensing element) exhibiting a memory effect, and also to an image sensor implementing said method.

[0002] [Prior art] A wide variety of image sensors exist, which vary according to the type of radiation to which they are sensitive and their mode of operation for capturing images.

[0003] In this document, radiation is understood to mean all types of external signals that originate from a scene and can reach an image sensor so that an image of the scene is captured by the image sensor. Each image takes the form of a set of intensity values ​​that are respectively assigned to points in the image and arranged in a matrix. In particular, radiation may be electromagnetic radiation in any wavelength range, in particular in the range of X-rays, ultraviolet, visible light, near-infrared and so-called thermal infrared. Radiation may also be acoustic radiation, in particular in the field of ultrasound.

[0004] The image sensor includes a matrix of sensitive elements, each capable of generating a detection signal that varies depending on the intensity of the radiation received by it, so that there may be a one-to-one correspondence between points in the image and the sensitive elements of the image sensor.

[0005] In the first type of sensor, the detection signal is acquired by exposing each sensitive element to radiation from the scene for a defined period (commonly called an integration period). The detection signal generated in the sensitive element by the radiation during the integration period is then read out by a dedicated circuit. The sensitive element is then reset (i) from the time of the detection signal readout, which is performed to capture an image, to (ii) the start of a new integration period for capturing a subsequent image. This reset ensures that images are captured consecutively corresponding to the individual integration periods. This specification does not relate to this first type of sensitive element or image sensor.

[0006] This specification relates to an image sensor in which sensitive elements can assume variable instantaneous states depending on the radiation received by the sensitive elements. In this second type of sensor, an image is captured by reading out values ​​specific to the instantaneous states of each sensitive element in the sensor matrix. In most cases, resetting each sensitive element is not possible or is not performed between two consecutive captured images. In this case, the detection signal read out from each sensitive element to capture an image depends not only on the radiation received by the sensitive element at the time of readout, but also on the radiation previously received by the sensitive element. In other words, the readout detection signal results from a combination of radiation intensities successively received by the sensitive element up to the time of readout. This effect is commonly referred to by those skilled in the art as the "memory effect." Due to this memory effect, each captured image results from a combination of successively occurring scene states. When the scene changes over time, the memory effect causes degradation of image quality. The memory effect can take different forms depending on the nature of the patterns in the scene and the movement (motion) of some of these patterns. In particular, the memory effect in the second type of image sensor can take the form of reduced image contrast, a tail effect that affects the movement of elements in the scene, etc.

[0007] A method has been developed to reduce the memory effect described above. The method involves applying an erasure sequence to at least one of the sensitive elements in the matrix of the image sensor between times when two successive images are captured. The erasure sequence is performed by a dedicated electronic circuit (sometimes called a thermalization circuit). The method is hardware-based in the sense that it acts on the sensitive elements in the image sensor.

[0008] Another method for reducing memory effects has been developed. This method is software-based and involves combining multiple raw images obtained directly as a result of successive readouts of the sensor elements. A matrix or multiple scalar coefficients are used to generate a temporal filtering function through convolution and / or linear combination. However, determining these coefficients is a difficult task, as they can cause a reduction in the information contained in the image and / or amplify image noise. Furthermore, the above-mentioned filtering operations are not adapted to correct for memory effects in the sensor elements within the image sensor. Therefore, the resulting correction is not optimized for memory effects.

[0009] [Technical issues] The object of the present invention is therefore to reduce image degradation caused by memory effects in the second type of image sensor described above, in particular by improving the transfer function of the image sensor and / or reducing tail effects that appear in the image when scene elements move.

[0010] [Summary of the Invention] To achieve this or another object, a first aspect of the invention proposes a method for capturing images, in which a number of images are successively captured using an identical matrix of sensitive elements, each of which exhibits a memory effect that makes the raw detection signal of said sensitive element dependent on the amount of radiation received by said sensitive element at the time of reading out said raw detection signal, and also dependent on the amount of radiation received by the same sensitive element prior to said time of reading out.

[0011] According to the invention, to form an image (called corrected image) at least partially corrected for the memory effect, each sensitive element of the matrix is ​​assigned an image point intensity value that is proportional to the difference between (i) the raw detection signal of said sensitive element read out for a newly captured image (called new raw image) and (ii) a portion of the raw detection signal of the same sensitive element read out for another image (called previous raw image) captured prior to the new raw image.

[0012] According to the above-mentioned correction, for each raw image constituted by raw detection signals read out from the sensitive elements, the image degradation caused by the memory effect of the sensitive elements is reduced in a corrected image constituted by image point intensity values. The previous raw image is used to estimate at least a part of the content of the memory effect, and this part of the content of the memory effect is subtracted (subtracted) from the content of the new raw image individually for each sensitive element. In this case, the resulting corrected image mainly corresponds to the radiation received by each sensitive element between two readout times of the plurality of raw images.

[0013] The correction of the memory effect described above involves replacing a raw image constituted by raw detection signals read out from the sensor element with an image constituted by image point intensity values ​​calculated according to the invention. The correction is even more effective if the "new" raw image and the "previous" raw image are captured in a short period between readouts of the sensor element. For this purpose, if multiple previous raw images are captured prior to the new raw image, the raw image used to calculate the image point intensity values ​​for the correction image may preferably be the last (newest) of the multiple previous raw images captured prior to the new raw image in the chronological order of capture of the raw images.

[0014] In a preferred embodiment of the present invention, the image point intensity value assigned to each sensitive element to form the corrected image may be proportional to the difference between (i) the raw detection signal of the sensitive element read out for the new raw image and (ii) the result of multiplying the raw detection signal of the same sensitive element read out for the previous raw image by exp(-Δ / τ), where exp(.) represents an exponential function, τ is the characteristic response time of the sensitive element, and Δ is the non-zero period between the readout times of the sensitive element for the new raw image and the previous raw image, respectively. This combination of the new raw image and the previous raw image to generate the corrected image is more appropriate when the behavior of each sensitive element (i) corresponds to a linear transfer function in time or (ii) can be approximately described by said linear transfer function in time. In the context of the present invention, the transfer function of a radiation sensitive element is denoted as f(s). The transfer function is the quotient of the Laplace transform of the intensity of the incident radiation with the Laplace transform of the raw detection signal generated by the sensitive element in response to the incident radiation. In other words, f(s)=A d (s) / A R (s), where s represents the Laplace variable, and f represents the transfer function of the sensor element. A R represents the Laplace transform of the intensity of the radiation incident on the sensitive element. drepresents the Laplace transform of the raw detection signal generated by the sensor. This transfer function is

number

[0015] In accordance with the preferred embodiment of the present invention described above, the image point intensity value assigned to each sensor element to form the corrected image may be proportional to the difference between (i) the raw detection signal of the sensor element read out for the new raw image and (ii) the raw detection signal of the same sensor element read out for the previous raw image multiplied by exp(-Δ / τ), divided by [1-exp(-Δ / τ)]. The division by [1-exp(-Δ / τ)] particularly avoids attenuation of the intensity of the corrected image, which may be even greater if the period Δ between the readout of the new raw image and the previous raw image is short. Optionally, an additional proportionality factor may be applied in addition to 1 / [1-exp(-Δ / τ)], thereby adjusting the scale of the image point intensity values ​​in the image corrected for memory effects. This additional proportionality factor may be constant, in the sense that it is independent of the period Δ between the readout of the new raw image and the previous raw image, and independent of the characteristic response time τ.

[0016] In various embodiments of the present invention, including the preferred embodiment described above, at least one of the following additional features may optionally be reproduced, either alone or by combining two or more of the features together:

[0017] The period Δ between the respective readout times of the new and previous raw images of the same sensor element may be shorter than the characteristic response time τ of this sensor element.

[0018] The period Δ between the respective readout times of a new raw image and a previous raw image of the same sensor element may be between 0.02 and 0.2 times the characteristic response time τ of this sensor element.

[0019] Each sensitive element may be a bolometer or microbolometer, a thermopile, a pyroelectric sensor, a ferroelectric sensor or a thermally deformable microlever sensor.

[0020] The sensitive elements in the matrix may be sensitive to electromagnetic radiation, including infrared radiation (especially thermal infrared radiation), X-rays, acoustic radiation (especially ultrasonic radiation), etc.

[0021] The same value of the characteristic response time τ may be common to all sensitive elements in the matrix.

[0022] The method may include a preliminary step in which a value of a characteristic response time τ is determined individually for each of the plurality of sensitive elements in the matrix, and the value of the characteristic response time τ thus determined for each sensitive element may then be used to calculate image point intensity values ​​derived from said characteristic response times to form a corrected image.

[0023] Each corrected image may be formed from a pair of raw images including (i) a new raw image and (ii) a previous raw image captured prior to the new raw image, where the pairs of raw images used to form successive corrected images may be disjoint, non-interlaced, and chronologically consecutive.

[0024] A number of corrected images may be formed by combining the same new raw image with a number of previous raw images captured consecutively before the new raw image. In this case, each corrected image is obtained by using, for a period Δ associated with each sensitive element, the difference between (i) the readout time of the sensitive element for the new raw image and (ii) the readout time of the sensitive element for the previous raw image used to form this corrected image. By comparing the corrected images thus obtained, a gradual reduction in the tail effect can be observed. In some cases, this reduction in the tail effect makes it possible to estimate the speed of movement of an element of the imaged scene when it is moving.

[0025] A second aspect of the present invention provides an image sensor comprising a matrix of multiple sensitive elements and an image processing module. Each sensitive element exhibits a memory effect that causes a raw detection signal read out from this sensitive element to depend not only on the amount of radiation received by the sensitive element at the time of readout of the raw detection signal, but also on the amount of radiation received by the sensitive element prior to the readout time. The image processing module outputs images at least partially corrected for the memory effect. Each corrected image is formed by image point intensity values ​​respectively assigned to the sensitive elements of the matrix. The image processing module calculates the image point intensity value of any one of the multiple sensitive elements as a value proportional to the difference between (i) the raw detection signal of the sensitive element read out for a new raw image and (ii) the raw detection signal of the same sensitive element read out for a previous raw image captured prior to the new raw image.

[0026] The image sensor may be adapted to perform the method according to the first aspect of the invention. Optionally, the image sensor may include a suitable mode of performance and any additional features.

[0027] [Brief description of the drawing] The configuration and advantages of the present invention will become more apparent in the following detailed description of some examples of non-limiting embodiments, with reference to the accompanying drawings, in which:

[0028] FIG. 1 is a block diagram of an image sensor according to the present invention; 2-4 show various sequences for generating a corrected image according to the present invention; Figure 5 shows the first effect of the present invention on image contrast; FIG. 6 illustrates a second effect of the present invention on tail effects that may appear in an image.

[0029] [Detailed Description of the Invention] According to FIG. 1, the image sensor 10 comprises a matrix 10a of a plurality of sensitive elements 1. The sensitive elements are arranged at the intersections of the rows and columns of the matrix and may be considered independent of one another. For example, the matrix 10a may consist of 320 x 240 sensitive elements 1. In particular, in a model known to those skilled in the art, all the sensitive elements 1 may be identical. For example, each sensitive element 1 may be a microbolometer. In this case, the sensitive element includes a section of conductive material whose electrical resistance varies as a function of temperature. This section of variable electrical resistance material is at least partially thermally insulated from its environment. Therefore, radiation R irradiating this section and being absorbed by it causes a temperature increase in this section, which in turn causes a change in its electrical resistance. This electrical resistance constitutes a raw detection signal that is read out corresponding to each captured image. This measurement principle is well known and will not be repeated here.

[0030] This type of microbolometer is also known to have a first-order transfer function in time, characterized by (i) the value of a gain coefficient, denoted G, and (ii) the value of a characteristic response time, denoted τ. This transfer function, denoted f(s) and dependent on the Laplace variable s, is given by:

number

number

[0031] To form the image sensor 10, the matrix 10a of sensitive elements 1 is associated with a controller 10b, designated CTRL. In a known manner, the controller 10b has (i) a power supply and readout function for each of the sensitive elements 1, (ii) an addressing function for each of the sensitive elements 1 in the matrix 10a, and (iii) possibly additional functions, such as a test function for the sensitive elements 1, an optional thermalization function for at least some of the sensitive elements, and a digitization function for the detected signals. In each readout cycle for all the sensitive elements 1 of the matrix 10a, the controller 10b outputs the values ​​of the detected signals read out from all the sensitive elements 1. In this case, one detected signal value is read out per sensitive element for each captured image. The reading out of these detected signals (referred to in the general part of this specification as raw detected signals) constitutes the image capture operation. The image obtained directly in this way is therefore referred to as a raw image. The readout of the matrix of sensitive elements may be carried out according to one of two modes: a rolling shutter mode or a snapshot mode. In the first case, the rows of the matrix are read out sequentially, one row at a time. In the second case, all rows are read out simultaneously. This difference in the readout mode of the image sensor does not affect the principles of the present invention or the results. The detailed description below is provided as a non-limiting example of the snapshot readout mode. To translate this description to the case of reading out an image using the rolling shutter mode, the readout times mentioned below can be considered to be the readout times of the first row of the matrix.

[0032] 1, an optical system may be used in front of the matrix 10a, which allows the scene to be imaged to be optically conjugate with the matrix of sensitive elements 1. In this case, the radiation R passes through this conjugate lens before it is incident on the sensitive elements 1.

[0033] In the present invention, the image sensor 10 further comprises an image processing module 11, which is connected to receive as input the raw image output by the controller 10b. The module 11 is designed to generate a processed image from the raw image, thereby making it possible to compensate for or at least partially correct the memory effect in the sensor element 1 mentioned above. For this reason, a processed image generated in accordance with the present invention by an image sensor including this module is referred to as a corrected image. The image processing module 11 may be a dedicated electronic circuit, a software module hosted in a processor, or a dedicated CPU.

[0034] According to the invention, the image processing module 11 generates a corrected image from a raw image by subtracting from the raw image another raw image that has been captured previously and multiplied by a defined coefficient. Multiplying a raw image by a coefficient is understood to mean an operation that includes multiplying all raw detection signals that make up the raw image by the coefficient. Furthermore, subtracting a first image from a second image is understood to mean an operation that includes calculating, independently for each sensor element 1, the difference between the detection signals read out for the second image and the detection signals read out for the first image. Therefore, according to the invention, S corr The corrected image marked (t) is

number

[0035] Instructively, the coefficient α (and possibly the coefficient β) is raw The readout time of (t-Δ) and the raw image S raw (t) may be chosen to vary as a function of the time period between the readout of the raw image S raw (t-Δ) and S raw (t) is the readout raw signal S i,j_raw (t-Δ) and S i,j_raw (t) is composed individually.

[0036] Optionally, at least one of the coefficients α and β may have different values ​​for distinguishable sensory elements 1 in the matrix 10a, in which case the values ​​of the coefficients α and β may be determined individually for each of the sensory elements 1 during a calibration or benchmarking step that may precede each image capture sequence or in a test laboratory.

[0037] In a preferred embodiment of the present invention,

number

[0038] In a further preferred embodiment of the present invention,

number

number

[0039] The above-described expression of α (and possibly β) makes it possible to effectively correct the memory effect when the sensitive element is of the type having a linear transfer function with respect to time, as mentioned above. Indeed, in this case, the raw image S, which is generally referred to as the previous raw image in this specification, raw When (t-Δ) is multiplied by α=exp(-Δ / τ), the raw image quantifies the contribution of the memory effect associated with all radiation received by each sensitive element prior to the capture of the raw image. This memory effect contribution, also referred to as the long-term memory effect, is used to calculate the new raw image S raw(t). According to the present invention, this memory effect contribution is corr In (t), the radiation intensity is completely removed. However, there remains a contribution from another memory effect, also called short-term memory effect, which is related to the radiation received by each sensitive element during the period Δ between the readout times of the two raw images.

[0040] If the sensitive element is not of the type having a linear transfer function with respect to time, the expression of the coefficient α as a function of the characteristic response time τ and the period Δ may still be used. Thus, an empirical value adapted to the sensitive element under consideration may be applied for the characteristic response time τ, even if this value does not have a theoretical significance related to the transfer function f(s) of the sensitive element.

[0041] Figure 2 shows a video sequence capture in which raw images are captured periodically at readout times spaced apart by a period Δ. In this case, the final raw image S raw (t) is the raw detection signal S at the read time t. i,j_raw (t) and the immediately preceding raw image S raw (t-Δ) is the raw detection signal S at the read time t-Δ i,j_raw (t-Δ) is composed of the raw image S preceding the raw image. raw (t-2Δ) is the raw detection signal S at the read time t-2Δ. i,j_raw (t-2Δ) is composed of another raw image S raw (t-3Δ) is the raw detection signal S at the read time t-3Δ i,j_raw (t-3Δ), and so on. In this case, a corrected video sequence made up of a number of images corrected according to the invention can be constructed at the same frame rate, equal to 1 / Δ, by combining each raw image with the last raw image read immediately before it. Therefore, the image point intensity values ​​S according to the invention can be calculated by i,j_corr (t) is the corrected image S corrTo obtain (t), we use the raw image S raw (t) and S raw (t-Δ) is combined with the image point intensity value S i,j_corr The corrected image S is composed of (t-Δ) corr To obtain (t-Δ), we use the raw image S raw (t-Δ) and S raw (t-2Δ) is combined with the image point intensity value S i,j_corr Corrected image S, which is composed of (t-2Δ) corr To obtain (t-2Δ), we use the raw image S raw (t-2Δ) and S raw (t-3Δ), and so on. The left axis in FIG. 2 corresponds to the time (point in time) coordinate, labeled t. For this purpose, the image sensor of the present invention may be equipped with a display system that is controlled to display the corrected image. In some cases, the display system is also controlled to display the raw image in accordance with the temporal correspondence. This allows evaluation of the improvement in image contrast and tail decay resulting from the correction of long-term memory effects. This mode of correcting images in a video sequence is particularly suitable for obtaining a sequence of corrected images at the maximum image capture rate possible in the matrix 10a and controller 10b.

[0042] Figure 3 corresponds to Figure 2 in the case where each corrected image is obtained by combining two consecutive raw images using pairs of disjoint, non-interlaced and consecutive raw images dedicated to different corrected images. i,j_raw (t) and S i,j_raw (t-Δ1) and the raw image S raw (t) and S raw (t-Δ1) is the corrected image S corr are combined to obtain the raw detection signal S i,j_corr (t-(Δ1+Δ2)) and S i,j_raw (t-(2Δ1+Δ2)) and the raw image S raw(t-(Δ1+Δ2)) and S raw (t-(2Δ1+Δ2)) is the image point intensity value S i,j_corr The corrected image S according to the present invention is composed of (t-(Δ1+Δ2)) corr The images in a video sequence are combined to obtain (t-(Δ1+Δ2)), and so on. Δ1 is the period between the readout times of both raw images of the same pair to obtain the corresponding corrected image. Δ2 is the period between the readout time of the previous raw image of the last pair and the readout time of the new raw image of the previous pair. The image frequency of the corrected images is therefore 1 / (Δ1+Δ2). This other mode of correcting images in a video sequence is suitable for obtaining images of the corrected images at low frame rates. This mode makes it possible to use a short period between the two raw images combined to obtain each corrected image.

[0043] Figure 4 also corresponds to Figure 2. In this case, we again consider a video sequence of raw images captured at a frequency of 1 / Δ. However, in this case, a series of corrected images is generated by progressively using the preceding raw images upstream in the video sequence, each time combining the same last captured raw image with a different raw image. For this reason, the image point intensity value S i,j_corr_1 The first corrected image S is composed of corr_1 is the raw detection signal S i,j_raw (t) and S i,j_raw (t-Δ) and the raw image S raw (t) and S raw (t-Δ) and

number

number

number

number

number

number

number

number

[0044] <Rendering image spatial frequencies in the presence of lateral motion> It is known that one way to emphasize the decay of the time transfer function of a matrix of sensitive elements is to capture an image with a spatial pattern that is periodic and moves with a constant velocity parallel to the direction of its periodicity. In this case, each sensitive element is subjected to radiation whose instantaneous intensity varies periodically according to a temporal frequency value equal to the product of its apparent velocity of movement and the period of the pattern. A scene is imaged on the matrix 10a of sensitive elements 1, consisting of bands parallel to the column direction of the matrix 10a and whose brightness varies sinusoidally parallel to the row direction of the matrix 10a. The spatial frequency of these bands in the scene image is ν s It is written as pixel -1 (pixels -1 ) The scene is moving at a constant speed parallel to the rows of the matrix 10a. V denotes the speed of movement of the image of the scene on the matrix 10a and can be expressed in units of pixels per second. All the sensitive elements 1 have the same value for the characteristic response time τ and their common transfer function is

number

[0045] <Tail effect> The matrix 10a used contains 320 columns and 240 rows of photosensitive elements 1. The characteristic response time τ of all photosensitive elements is approximately 14 ms. The image sensor consisting of this matrix of photosensitive elements captures an image sequence with a uniform background consisting of a black body at 325 K (Kelvin) through the top of an opaque rotating disk with three radial slit apertures. The matrix 10a is optically conjugate with the rotating disk, whose rotation axis is parallel to the optical axis of the conjugate optical system used. The rotation speed of the disk is 1.5 revolutions per second. The raw image acquisition rate is 60 images per second, corresponding to the period between the respective readout times of any two consecutive raw images (equal to 16.7 ms). The diagram in Figure 6 shows the S raw (t-Δ) and S raw The diagram represents the values ​​of the raw detection signals read out in the rows of the matrix 10a for two raw images in a video captured successively, each denoted as (t). raw (t-Δ) and S raw (t) is the corrected image S obtained by the present invention. corr For (t), the image point intensity values ​​for the same row of the matrix 10a are also shown. The movement of one of the radial slits in the rotating disk is visible between both raw images, from right to left in the diagram. The horizontal axis in the diagram of FIG. 6 represents the number of sensitive elements 1 in the row considered in the matrix 10a, and the number of columns n. c The vertical axis is the number of images S raw (t), S raw (t-Δ), and S corr For (t), we denote the value of the raw detected signal or intensity value of an image point, I, and the raw image, S raw (t) and S raw The gradually sloping rising and falling edges in the intensity curve, corresponding to the two edges of the slit at (t-Δ), constitute a tail effect due to a combination of the movement of the slit and the memory effect of the sensitive element 1.corr In (t), the two edges of the slit are more clearly visible, demonstrating the effectiveness of the memory effect correction. In this example, the formulas provided herein above for the coefficients α and β are:

number

[0046] It is understood that the invention can be reproduced while modifying the second aspect of the embodiment described in detail above, while retaining at least some of the cited advantages. In particular, the selection of raw images to be combined in pairs to obtain a corrected image can be modified with respect to the example shown. Furthermore, the use of a factor β that depends on the period Δ between the combined raw images will be considered optional, even if the factor α used is calculated according to this period Δ. [Brief explanation of the drawings]

[0047] [Figure 1] 1 is a block diagram of an image sensor according to the present invention; [Figure 2] 3A-3D illustrate various sequences for generating a corrected image according to the present invention. [Figure 3] 3A-3D illustrate various sequences for generating a corrected image according to the present invention. [Figure 4] 3A-3D illustrate various sequences for generating a corrected image according to the present invention. [Figure 5] 1 shows a first effect of the present invention on image contrast. [Figure 6]10 shows a second effect of the present invention on the tail effect that may appear in an image.

Claims

1. 1. A method for capturing an image, comprising: A plurality of images are successively captured using the same matrix (10a) of sensitive elements (1), The raw detection signal (S ｉ，ｊ＿ｒａｗ each sensitive element exhibits a memory effect that makes the raw detection signal (t) dependent not only on the amount of radiation received by the sensitive element at the time of reading out the raw detection signal, but also on the amount of radiation received by the sensitive element prior to the time of reading out the raw detection signal, A corrected image (S) at least partially corrected for the memory effect. ｃｏｒｒ Each sensitive element (1) of said matrix (10a) is assigned an image point intensity value (S ｉ，ｊ＿ｃｏｒｒ (t)) are individually assigned, The corrected image (S ｃｏｒｒ The image point intensity values ​​(S) assigned to each sensitive element (1) are used to form the image point intensity value (S ｉ，ｊ＿ｃｏｒｒ (t)) is (i) The raw detection signal (S raw (t)) of the sensor element read out for a newly captured image, called the new raw image (S raw (t)). ｉ，ｊ＿ｒａｗ (t)) and (ii) The raw detection signal (S raw (t−Δ)) of the same sensor element read out for another image, called the previous raw image (S raw (t−Δ)), captured prior to the new raw image. ｉ，ｊ＿ｒａｗ (t-Δ)) multiplied by exp(-Δ / τ), is proportional to the result of dividing the difference between exp(.) represents the exponential function, τ is the characteristic response time of the sensor element; A method wherein Δ is a non-zero period between readout times of the sensitive elements for each of the new and previous raw images.

2. The new raw image (S ｒａｗ (t)) and the preceding raw image (S ｒａｗ 2. The method of claim 1, wherein the period Δ between each readout time of a same sensitive element (1) for each of (t-Δ) is shorter than the characteristic response time τ of said sensitive element.

3. The preceding raw image (S ｒａｗ (t-Δ)) is the time series of capturing the raw images. ｒａｗ 3. The method of claim 1, wherein the image is the final image of a plurality of images captured prior to (t).

4. 4. The method according to any one of claims 1 to 3, wherein each sensitive element (1) is a bolometer or microbolometer, a thermopile, a pyroelectric sensor, a ferroelectric sensor or a thermally deformable microlever sensor.

5. Each corrected image (S ｃｏｒｒ (t), S ｃｏｒｒ (t-(Δ １ +Δ ２ ))) is formed from a pair of raw images, The above pair of raw images are (i) New raw image (S ｒａｗ (t), S ｒａｗ (t-(Δ １ +Δ ２ )))and, (ii) A preceding raw image (S ｒａｗ (t-Δ １ ), S ｒａｗ (t-(2Δ １ +Δ ２ ))) and, 5. A method according to claim 1, wherein the pairs of raw images used to form successive corrected images contain no common elements, are non-interlaced, and are chronologically consecutive.

6. A plurality of corrected images (S ｃｏｒｒ＿１ , S ｃｏｒｒ＿２ , S ｃｏｒｒ＿３ , S ｃｏｒｒ＿４ )teeth, The same new raw image (S ｒａｗ (t)) is obtained by dividing the new raw image by a plurality of preceding raw images (S ｒａｗ (t-Δ), S ｒａｗ (t-2Δ), S ｒａｗ (t-3Δ), S ｒａｗ (t-4Δ)), 5. A method according to any one of claims 1 to 4, wherein each corrected image is obtained by using, for the period Δ associated with each sensor element (1), the difference between (i) the readout time of the sensor element for the new raw image and (ii) the readout time of the sensor element for the previous raw image with which it is combined to form the corrected image.

7. An image sensor (10), a matrix (10a) of a plurality of sensitive elements (1); an image processing module (11); The raw detection signal (S ｉ，ｊ＿ｒａｗ each sensitive element exhibits a memory effect that makes the raw detection signal (t) dependent not only on the amount of radiation received by the sensitive element at the time of reading out the raw detection signal, but also on the amount of radiation received by the sensitive element prior to the time of reading out the raw detection signal, The image processing module generates a corrected image (S ｃｏｒｒ (t)) and outputting an image Each corrected image is calculated by dividing the image point intensity values ​​(S) respectively assigned to the sensitive elements (1) of the matrix (10a). ｉ，ｊ＿ｃｏｒｒ (t)), The corrected image (S ｃｏｒｒ The image point intensity values ​​(S) assigned to each sensitive element (1) are used to form the image point intensity value (S ｉ，ｊ＿ｃｏｒｒ (t)) is (i) The raw detection signal (S raw (t)) of the sensor element read out for a newly captured image, called the new raw image (S raw (t)). ｉ，ｊ＿ｒａｗ (t)) and (ii) The raw detection signal (S raw (t−Δ)) of the same sensor element read out for another image, called the previous raw image (S raw (t−Δ)), captured prior to the new raw image. ｉ，ｊ＿ｒａｗ (t-Δ)) multiplied by exp(-Δ / τ), is proportional to the result of dividing the difference between exp(.) represents the exponential function, τ is the characteristic response time of the sensor element; Δ is a non-zero period between readout times of the sensitive elements for each of the new and previous raw images.

8. An image sensor (10) according to claim 7, adapted to carry out a method according to any one of claims 1 to 6.

Images