RU212327U1 - HIGH DYNAMIC RANGE IMAGE PRODUCER - Google Patents

Abstract

The present utility model relates to a high dynamic range imaging device. The technical result of using the utility model is to increase the accuracy of solving the problem of calculating the direction vector to the sighted objects of the scene, for example, calculating the direction vector to the Moon relative to the stars due to the fact that images are obtained with a single exposure time of all objects in the scene. A device comprising a lens for exposure of the photosensitive matrix, a memory means for storing image data and a processor capable of processing image data, according to the utility model, the photosensitive matrix is configured to provide simultaneous acquisition of two images with the same exposure time, of which one image is optimized for viewing bright objects low-gain scenes, and of which the second image is optimized to view the faint objects of the high-gain scene. The processor of the device executes the algorithm for processing the array of pixel intensities of the first image obtained with a low gain Y _H, and the array of pixel intensities of the second image obtained with a high gain Y _B, and calculates the pixel intensities of the final image with a high dynamic range Y _SDD . The device can be most effectively used to create high-precision, small-sized, low-power instruments for orientation and navigation systems of small and microspacecraft using astronavigation algorithms. 2 ill.

Description

Technical field

The present utility model relates to a device for creating images with a high dynamic range, which can be used in vision systems to solve the problems of selecting objects with very different brightness and outputting a direction vector to these objects with high accuracy.

The device can be most effectively used in technical vision systems, in particular, in spacecraft control and navigation systems in solving problems of automatic orbital servicing.

Prior Art

It is known that the composition of various means of orientation and navigation of spacecraft includes devices for stellar orientation. Typically, stellar orientation instruments include a housing; an optical scheme consisting of a lens and a matrix photodetector; an electronic block consisting of a calculator, an electronic memory containing a star catalog. Star orientation devices determine the orientation parameters by comparing the image of a section of the starry sky with a star catalog, (see Dyatlov S.A. Review of spacecraft orientation sensors Dyatlov S.A. et al. // All-Russian scientific and technical conference "Modern problems of orientation and navigation of space devices ". - M. - IKI RAS. - 2008, - pp. 11-32). In fact, a star orientation device is a video camera designed to obtain images of informative parts of the celestial sphere. Further processing of the obtained images by means of the electronic unit and the implementation of special algorithms make it possible to ensure that the requirements of the assignment are met. The successful solution of the tasks set largely depends on the quality of the images obtained.

When solving problems of orientation and navigation of space vehicles, it often becomes necessary to obtain images of astronomical objects, such as the Moon, against the background of stars. The difficulty in obtaining high-quality images lies in the fact that the stars and the Moon have a very different level of brightness: the brightness of the Moon is minus 12.8 magnitudes, while the brightness of one of the brightest stars, Sirius, is minus 1.47 magnitudes. In this case, to solve the orientation problem, an image of at least 10 stars with a brightness of up to +6 magnitude must be obtained. According to Norman Pogson's formula, it can be calculated that the brightness of the Moon can exceed the brightness of the dimmest object of the viewed scene by 15,000,000 times.

Obtaining an image of a scene containing objects with such a spread in brightness in one frame using conventional cameras leads to the fact that the dynamic range of the photosensitive matrix is not enough to obtain a high-quality image. If you set the exposure time to get a clear image of a bright object, dim objects will be underexposed and may be completely absent from the image. In the case of setting a long exposure time, sufficient to fix dim objects, bright objects will be perlite, will not have clear boundaries and will not be designed to solve the problem of calculating the direction vector to these objects. Fixing the image of such objects with a quality sufficient for their recognition, selection and calculation of the direction vector relative to the optical coordinate system is possible using a device with a high dynamic range corresponding to the brightness range of the objects of the viewed scene.

To obtain high-quality images with a high dynamic range, a device is known in which a multilayer module consisting of several charge-coupled devices (CCDs) is used as a photosensitive matrix (patent RU 2012156 C1, published 20.04.1994). Also in the known device, light-protected sections for storing accumulated charges are implemented. During the accumulation of charges by means of the light-sensitive surface of the CCD arrays, these charges are repeatedly transferred to the light-protected sections and summed up in them. The high dynamic range is achieved by the fact that the depth of the potential well of the light-protected sections is greater than that of the CCD matrices, thereby making it possible to accumulate a charge with a greater intensity.

Devices are also known in which a high dynamic range is ensured by the fact that the resulting image is composed of several images obtained with different exposures. A similar approach is implemented in the Astrol-17 star sensor developed by JSC NPO Lepton. The main purpose of the device is to determine the astroorientation of the associated coordinate system relative to the inertial coordinate system by the stars. This device also implements the function of issuing target designation (direction vector) to an object located in the field of view of a device that is not a star. To obtain images of objects with a brightness exceeding the brightness of the sighted stars, the principle of multiple exposure is used. One scene is exhibited at least three times with different exposures.

As a prototype for the claimed utility model, a device for creating images with a high dynamic range from multiple exposures is considered (patent RU 2397542 C2, published on January 31, 2006). The prototype implements the following principle of operation. Light is obtained by making a series of first exposures in conjunction with one second exposure in rapid succession. The first exposures are combined to provide sufficient brightness and sharpness. The second exposure is for collecting color information. The first exposures are combined with the second exposure to provide an enhanced image.

The disadvantage of the prototype is that the final image is obtained by exposing several frames, which are obtained at different points in time. Due to the exposure time shift, the positions of objects on different frames will also have a shift relative to the optical coordinate system of the device, which will inevitably introduce an error into the solution of the final problem in the case of using the prototype in vision systems designed to calculate the relative position of sighted objects.

Utility Model Disclosure

The technical result of using the utility model is to increase the accuracy of solving the problem of calculating the direction vector for sighted scene objects, for example, calculating the direction vector to the Moon relative to the stars, due to the fact that images are obtained with a single exposure time for all scene objects.

The specified technical result is achieved by the fact that a device for creating an image with a high dynamic range, containing a lens for exposure of the photosensitive matrix, a memory means for storing image data, and a processor capable of processing image data, according to the utility model, the photosensitive matrix is configured to provide simultaneous acquisition of two images with the same exposure time, of which one image is optimized for sighting of bright scene objects, with low gain, and of which the second image is optimized for sighting of dim objects of the scene with high gain, and that the processor of the device provides the processing algorithm for processing an array of pixel intensities the first image obtained with a low gain Y _H and the array of pixel intensities of the second image obtained with a high gain Y _B in the computing processor of the casting device and calculating the pixel intensity of the final image with a high dynamic range Y _SHDD depending on Y _B and Y _H , while the pixel intensities of the image obtained with a low gain Y _H are recalculated taking into account the scale factor and the intensity compensation factor.

It is essential to achieve the technical result that the utility model uses a photosensitive matrix that has two information outputs: output 1 and output 2. When viewing the scene, an image is obtained from each of the outputs simultaneously, with one exposure time, but different analog amplification , where the low gain image is optimized for bright objects in the scene, and the high gain image is optimized for low sensitivity and dim objects in the scene. The user can combine high and low gain images to create one final high dynamic range image using an external algorithm stored in the compute unit's memory.

A new set of essential features has been created, which provides a technical result.

Brief description of the drawings

In FIG. 1 shows a block diagram of the utility model. The main elements are: a lens 1 designed to expose a photosensitive matrix 2, a computing unit 3 which includes a processor 4 designed to implement operation algorithms and a memory 5 designed to store image data and algorithms.

In FIG. 2 shows a block diagram of the algorithm for generating the final image with a high dynamic range, which is implemented in the proposed utility model.

Implementation of the utility model

The utility model relates to a device that is designed to solve the problem of selecting objects with very different brightness and issuing a direction vector to them with high accuracy relative to the optical coordinate system of the device.

Apparatus for creating a high dynamic range image FIG. 1, containing a lens 1 for exposure of a photosensitive matrix 2, a computing unit 3 which includes a processor 4 designed to implement the operation algorithms and a memory 5 designed to store image data and algorithms. The processor processes the image data in accordance with the algorithm in FIG. 2 and forms the final image containing the image of the stars and the Moon with a single exposure time and suitable for calculating the direction vector to the Moon relative to the stars with high accuracy.

According to the algorithm in Fig. 2 provides processing of the array of pixel intensities of the first image obtained with low gain Y _H and the array of pixel intensities of the second image obtained with high gain Y _B in the processor of the computing device and calculating the pixel intensity of the final image with high dynamic range Y _FDD depending on Y _B and Y _H , while the intensities of the image pixels obtained with a low gain Y _H are recalculated taking into account the scale factor and the intensity compensation factor.

The utility model is implemented on the basis of software and hardware of the Astrol-17 star tracker. The characteristics of the lens, processor and memory of this star tracker are suitable for the implementation of the proposed utility model. These components will not be described in this application. To implement the utility model, instead of the Kodak KAI 2020 photosensitive matrix, it is necessary to use a photosensitive matrix that provides two images exposed at the same time, but with different gain factors: low gain - K _H , which is optimized for sighting bright objects in the scene , and with a high gain - K _B , which is optimized for sighting dim scene objects. Currently, such photosensitive sensors are on the market, for example, the GSENSE line of sensors from Gpixel. The matrices are made using CMOS technology and are produced with a resolution of 2048 × 2048 pixels. The power consumption of matrices does not exceed 1.5 W. Also, the devices of this family are produced in the version with "back-illumination", which provides a quantum efficiency of at least 85%. The operating temperature range is from minus 55°С to +85°С (see electronic source, access mode - free http://image-sensors-world.blogspot.com/2019/12/gpixel-gsense2020bsi-scmos-spec. html). Both of these matrices: Kodak KAI 2020 and matrices from the Gpixel GSENSE line have a similar output video interface, which will allow the matrix to be replaced with minor modifications to the device.

To form the final image from two images obtained with the same exposure time, but with different gains K _H and K _B , which are optimized to obtain, respectively, the brightest and dimmest object on the viewed scene, an algorithm is implemented in the device processor in accordance with the block the diagram shown in Fig. 2. Images from the photosensitive matrix outputs enter the computing unit in the form of two pixel intensity arrays Y _{H i,j} and Y _{B i,j} , where i=0…N, j=0…M are the number of rows and columns of the photosensitive matrix, respectively. The pixel intensity of the final image with a high dynamic range (Y _{VDD i,j} ) is formed in accordance with the following relationship

- масштабный коэффициент;

- коэффициент компенсации интенсивности; L_B - смещение уровня черного относительно изображения, полученного с высоким коэффициентом усиления; L_H - смещение уровня черного относительно изображения, полученного с низким коэффициентом усиления.where T _H is the threshold value of the intensity of the pixels. Corresponds to the minimum value of the pixel intensity of the image obtained with low gain;

- scale factor;

- intensity compensation factor; L _B is the offset of the black level relative to the image obtained with a high gain; L _H is the offset of the black level relative to the low gain image.

The utility model makes it possible to obtain clear images of objects with a high dynamic range, with a quality sufficient for recognizing and selecting objects and issuing target indications on them relative to the optical coordinate system. A high dynamic range is selected in advance by selecting the gain factors K _H and K _B depending on the tasks being solved and the possible spread in the range of brightness of objects in the viewed scene. The coefficients are stored in the device's memory. The technical result of using the utility model is to increase the accuracy of solving the problem of calculating the direction vector for sighted scene objects, for example, calculating the direction vector to the Moon relative to the stars, due to the fact that images are obtained with a single exposure time for all scene objects. The device can be most effectively used to create high-precision, small-sized, low-power instruments for orientation and navigation systems of small and microspacecraft using astronavigation algorithms.

Claims (1)

A device for creating images with a high dynamic range, containing a lens for exposure of the photosensitive array, a memory means for storing image data and a processor capable of processing image data, characterized in that the photosensitive array is configured to provide simultaneous acquisition of two images with the same exposure time, from wherein one image is optimized for viewing bright objects in the scene at low gain, and of which the second image is optimized for viewing dim objects in the scene at high gain; the device processor implements an algorithm for processing the array of pixel intensities of the first image obtained with a low gain Y _H, and the array of pixel intensities of the second image obtained with a high gain Y _B, and calculating the pixel intensity of the final image with a high dynamic range Y _FDD depending on Y _B and Y _H , while the intensities of the image pixels obtained with a low gain Y _H are recalculated taking into account the scale factor and the intensity compensation factor.

Images