DE3630739C1 - Method for data pick-up by means of detector arrays and devices for carrying out the methods - Google Patents

Abstract

To pick up data by means of a detector array, a selective change of location is carried out between object and array which provides a higher resolution of objects. This makes it possible to achieve signal improvements by averaging (array and object noise), colour cameras or high-resolution structures. <IMAGE>

Description

The present invention relates to methods and devices for data recording, in which an object is wholly or partly open a detector array is imaged and those generated by it Signals are processed further.

Obtaining analog and / or digital measurement results of objects has long been known, the objects either by a single sensor (scan) or by one regular arrangement of many sensors (detector array) detected will. An example of this is the acquisition of the gray values of photographic images by microdensitometer or by CCD camera as an area sensor. When using these devices fluctuations in the measured values occur, which are both device-specific, and may also be due to the submission.

In the case of area sensors, the individual sensor elements are included same external signal influence more or less different Output signals. These signal differences despite Uniform external signal effects can be achieved by multiple or longer measurement at the same location with the same, immobile sensor element can not be reduced. A single pixel correction, as described in US Pat. No. 4,317,134, by multiplying the signals from the individual sensors by predefined correction factors from a PROM table can be found here to a more even sensor signal output after signal processing to lead. The greater the number of sensors on a detector array but is, the greater the times become for that Correction. The time problems that arise can be if at all, they are difficult to reduce and limit Number of recordings possible.

When shooting objects with strong signal differences occur on a small area when using detector arrays Problems due to the fact that there is between the recording sensors  free areas are located which do not have a signal for the recording deliver. You have this problem especially when presenting photographic images. In these, fluctuations in the Grayscale signals through the grain and its statistical distribution in the photographic emulsion. These Gray value fluctuations are inversely proportional to the diameter the measuring surface. Through multiple or longer measurements In the same place, this noise cannot be reduced will. A method is described in EP-A1-1 20 678 which a CCD sensor vibrates in a controlled manner to eliminate color moir effects. This procedure is suitable to reduce the grain noise by the sensor at the picture over a certain area in a shaky movement to be led. The sensor is thus during a recording therefore moves and takes up a larger area. Thereby finds an unweighted averaging of the recorded areas instead of. With the very precise work for surveying tasks, how they occur in the field of photogrammetry have an effect but the high susceptibility to frequency stability and the device-related sensitivity of the sensor vibration to external vibrations (e.g. road, rail traffic) very much negative from what is generally restricting the use or the achievable reproducibility of the Measurement affects.

From DE-PS 6 77 773 is known, by means of two pairs of rotary wedges a mechanical image decomposition or image construction device for To build television purposes, one side Movement of a slit image behind a rotating one Slotted disc is made. This special construction for For television purposes, however, data recording is not suitable to improve a detector array in which the exposure done simultaneously on the entire array.

From GB-PS 20 85 629 and DE-PS 30 05 206 it is known take several measurements or recordings of an object and to base this on further processing. In particular is known from GB-PS 20 85 629, several cameras in one  Arrange semicircle around a moving object on a Assembly line can also be viewed from the side.

The invention has for its object methods for data acquisition by means of detector arrays that are characterized by Signal improvement properties in terms of noise reduction distinguish and devices for performing this Creating procedures.

The methods according to the invention are the subject of the claims 1 to 7, while claims 8 to 13 the structure of Relate to devices for performing these methods. In addition, the invention allows a much higher one Achieve resolution on the detector arrays.

The method according to claim 1 is characterized in that a change in existing devices, such as, for example, is not necessary here. B. the Zeiss PLANICOMP, must be done to achieve a signal improvement. This method can be used with all devices equipped with an xy table and allows an immediate improvement of the image recording, both with regard to the suppression of the grain noise and the reduction of the sensor noise. This double reduction in noise is achieved solely by multiple measurements, which are linked to one another by means of an averaging, the object being moved successively into individual recording positions. These individual recording positions differ by preselectable amounts; depending on how strong the noise of the original and / or the sensor is and what magnification you are working with. After the measurements for a surface element of the object have been completed, the object is moved to the next surface element to be recorded. The change in location of the object necessary to accommodate a surface element is carried out periodically. Another major advantage of the method is the precision that can be achieved by using stepper motors for object movement.

A further improvement in terms of array noise can be achieved if the array is regularly irradiated during the object movement to determine current, characteristic values. By means of this white keying, the different recording sensitivity of the sensors can be determined. If you are dealing with an action on the sensor that can be converted into a straight line using mathematical methods, the sensitivity line is characterized by the following equation S = c · E + d . The sensor signal S is thus obtained when the product of the intensity of the action E and the amplification factor c is added to the offset d . If you now want to correct the sensor, a white key generates an exact determination only for an intensity of the action E. All other values are only approximate. An exact correction is only possible if the offset and the gain factor are determined. This can be done by a further white keying, the intensity of this sensor action should be as far as possible from the intensity of the previous sensor action in order to be able to determine the compensation values as precisely as possible. This is the case when the array is read out if there is no external influence on the sensors (black keying). In the case of arrays whose sensitivity curve cannot be expressed by a straight line, an approximation in the required quality can be carried out by measuring several intensities. Because this correction is carried out regularly, a change in the sensitivity of the detector array over time is automatically taken into account. This is a major advantage over previously used methods. A corresponding signal delay ensures that the useful signals are available at the exact time with the correction signals for an output value correction of the detector array. The values of the black and white keying can also be used as the identity code of the arrays, e.g. B. to prevent third-party video recordings in security systems.

Now you make sure that there is between the array and the optical system a moving filter with multiple sectors is attached, so you can do certain operations by changing the sectors the shooting position (e.g. color separations). The flexibility the system is expanded very much because the  Number of sectors no limits.

The method according to claim 2 is characterized in that by the rotational movement of at least one detector array rapid change of location is feasible and the signal recording takes place in a discrete, stationary manner. The hereby resulting possibility, only through the larger movements Carry out changes of location between object and recording unit to have to, being both the movement of the object and the movement of the recording unit can also be selected, allows small periodic changes of location very much faster than before. The recording with the vibration movement requires shorter total recording times, works but not as exact. The rotary motion will relate to the object from the detector array has several selectable recording positions one after the other, discreetly and in recurring order ingested.

The change of location by rotary motion can be done by several methods be realized. The detector array becomes vertical rotated to the optical axis, the array can be on a drum be attached. This drum rotates the array after one certain time back in a fixed recording position. He follows now an axial back and forth movement during the rotation of the drum this drum, it has between the starting position of the Drum and the drum position after one revolution a certain change of location precisely preselectable by the mechanism experienced in the direction of the axis of rotation. These can be selected precisely axial change of location may take several revolutions before it is undone by the mechanism. On Several arrays can also be attached to the drum be brought into the receiving position one after the other.

Instead of the drum movement in the axial direction, the detector arrays can also be arranged on the drum with a slight offset. If the drum is now rotated, the recording array moves back and forth in the axial direction depending on the current position of the drum in accordance with the detector arrays attached to it. One detector array each is in the recording position, while the or the unused array is read out and, if necessary, white / black keyed. This local movement between two recordings of the object can be used via multiple readings to obtain a noise reduction by averaging. Depending on how many images are to be used for averaging, the drum can be provided for n detector arrays. A combination of the axial drum movement and the slight displacement of the arrays on the drum is possible.

However, the detector arrays cannot only work on one drum attached to be rotated. A turntable can also be used as a bracket serve for the array. Here is the axis of rotation parallel to the optical axis. If you set the axis of rotation exactly through the geometric center of the array receiving surface, so becomes the same unit of area of the object with each shot mapped onto the array. Depending on the pattern of the sensors on the Arrays are in the 90 ° -180 ° -270 ° -360 ° position other points of the object are shown. The recordings can be used for an averaging by appropriate linkage will. Only in the special case of square room division the sensors on the array become exactly the same point mapped on the array and the averaging of the different Shooting from the shooting positions causes a reduction of sensor noise. On the other hand, you have a rectangular distribution pattern of the sensors on the array takes place when they are linked a decrease of the four shots at two positions of sensor noise while the other two shots can be used for signal enhancement of the template. Other distribution patterns on the array can only be for one Use signal enhancement of the template to be scanned. Lays you consciously turn the axis of rotation outside the geometric center, so are other distribution patterns of point detections achievable if you link several recordings together. Do you have detector arrays that are not rectangular in shape? have the corresponding other angular positions to use.  

The method according to claim 3, however, is characterized that the object is improved by means of a adjustable between the detector array and the object optical device for deflecting the coming from the object Radiate successively over preselectable amounts at small angles is moved over the detector array, multiple measurements for an averaging can be made.

One possible component that can be used is the wedge. For him, the approximate value for a small angle of incidence α is a deflection angle δ of

δ = - α (n -1)

with n - refractive index.

The wedge can be corrected achromatically in order to eliminate disturbing color separations of the amount δ / γ (γ- Abbe number) for color images. By inserting the wedge between the object and detector array and rotating or translational movements of the wedge, the small change in location required between two exposures can now be made. The multiple measurements obtained in this way can be used for a message.

The change of location carried out using the versions described above can be exploited in several ways. The detector array can put a suitable mask in front of his shooting position or filters for generating color images or for image processing purposes to have. In the case of several arrays, corresponding ones can Filters are located directly in front of an array. The Filters between detector array and object can be moved Sector filters exist. To improve the image, a White / black keying takes place, as under explanation of the The method according to claim 1 described. To be particularly advantageous can be seen that the construction of a color camera with only one array is possible and besides the rotating array or filter no changes to the hardware and just minor software adjustments are necessary. One uses the individual shots from the different positions of the Sensor with respect to the object not for the generation of color images,  so is an improvement in the resolution of the array possible, the positions in the inactive spaces now scanned on the receiving surface (conductor tracks, memory, etc.) will. The processing of these recordings is linked with computing systems with a parallel data processing structure can be particularly advantageous because these computers can each calculate the data of a recording. So these computers can already use the existing software be used for this purpose, slight software adjustments provided.

The invention is explained below with reference to FIGS. 1-13.

Fig. 1 Enlargement of an array with dimensions;

Fig. 2 is a chip array in detail;

Fig. 3 is a collection assembly comprising xy roulette table;

FIGS. 4 and 5 is a two-chip solution of rotation from two sides;

Fig. 6 is a multi-chip rotary embodiment;

Fig. 7 is a rotating recording arrangement on a turntable;

Fig. 8 is an illustrative example of the hub assembly;

Fig. 9, a solution of the optical deflection;

Fig. 10 shows a further optical deflection means of a wedge Abat'schen;

FIG. 11 is an optical deflection means of a vibrating wedge pair;

FIG. 12 is an optical deflection means of a rotary wedge pair principle;

Fig. 13 shows a practical embodiment of the Fig. 12.

Fig. 1 shows in detail a cross section of a detector array for explaining the general averaging method. When using photographic images as a template, the noise caused by the photographic grain is reduced by multiple measurements with a slightly shifted image. Multiple measurements with certain changes in location between the detector array and the object both reduce the grain noise and the sensor noise caused by the different sensitivity of the sensor elements. The changes in location can be recorded mathematically as follows:

V = V _{x 2 x 1} + _x n · _x a + m _x b · _x

and

V _y = V _{y 2 y 1} + n · a · _y + m _y b _y

With

V ₁- image recording position 1 V ₂- image recording position 2 a _{x, y} - sensor pixel length b _{x, y} - sensor pixel active surface length

In one exemplary embodiment, shown in FIG. 2, two measurements are made for each image zone, which are each shifted in the xy direction by the same amount. The number of measurements for the averaging and the optimal change of location must be adapted to the respective object and array. In FIG. 2, the individual sensors of an area array ( 10 ) in the 1st measurement are designated with capital letters, and the sensors of the same area array in the 2nd measurement with lower case letters. During the evaluation, the mean value of the signal from element aa of the second measurement is formed with the signal from element CC of the first measurement, generally the mean value from the signals of elements i, k of the first measurement with elements i + n _x + m _x , k + n _y + m _{y of} the 2nd measurement.

FIG. 3 shows an embodiment in the manner of method 1 for changing the location of the object when the picture is taken. Here, the desired averaging per pixel is achieved by adding multiple measurements in a computer, the image shifting being carried out by moving the template on the evaluation device. The image ( 1 ) to be digitized is attached to an image carriage ( 2 ) (with a transparent plate). This image carriage ( 2 ) can be moved in one direction and can be precisely positioned with a motor-driven spindle ( 3 ) by means of a stepper motor ( 11 ). An intermediate carriage ( 4 ), which is also driven by a stepper motor ( 12 ) via a spindle ( 5 ), is used for positioning in the direction y perpendicular thereto. The object to be recorded is illuminated by a light source ( 6 ) with appropriate optics ( 7 ). An image section ( 8 ) of the overall image ( 1 ) is imaged on an area sensor ( 10 ) by means of optics ( 9 ), which consists of a matrix of individual sensors (detector array) (e.g. CCD sensor). For multiple scanning with a shifted image, the image carriage ( 2 ) and / or the intermediate carriage ( 4 ) is now moved into a new recording position either in the x direction and / or in the y direction

V ₂ = n × a + m × b + V ₁

With

a = (a _x , a _y ), b = (b _x , b _y ), n = (n _x , n _y ), m = (m _x , m _y ) V ₁ = (V _{2 x} , V _{2 y} ) and V ₁ = (V _{1 x} , V _{1 y} )

postponed.

Fig. 4 now shows a recording arrangement according to claim 5 with two detector arrays on a rotatable drum ( 13 ) with the axis of rotation ( 14 ). One array ( 18, 19 ) is in the receiving position, while the other is read out and the array-specific characteristic values can be determined. The first recording is made by the array 19 . After the recording has been completed, the array ( 18 ) is positioned by turning ( 16 ) or clockwise ( 17 ) about the axis of rotation ( 14 ) of the drum ( 13 ). While one array ( 18 ) now provides a picture of the object, the other array ( 19 ) can determine array-specific characteristic values with which the picture of array 19 is corrected. By further ( 15 ) or backward rotation ( 16, 17 ) of the drum ( 13 ), the array ( 19 ) can then be brought back into the receiving position, while the other array ( 18 ) is now read out and the array-specific values are determined. If the two arrays are slightly offset on the drum, the object has apparently moved to a different location after a rotation and multiple readings can reduce the noise. Depending on the construction used, the xy positioning of the object to be photographed or the drum takes place. It should be borne in mind that the drum with the two arrays only replaces the array ( 10 ) in FIG. 3. If the arrays are not to be used for image acquisition on an xy table, a corresponding other periphery must be selected for the entire recording arrangement. In FIG. 5, the arrangement just described is rotated by 90 °. If, instead of displacing the axis of rotation ( 14 ) in the direction of an array on the drum, the entire drum is axially displaced in the direction of the axis of rotation ( 14 ) by means of a mechanism, the array can move in the receiving position and one has the advantage that you can do without a second array.

Fig. 6 shows a further embodiment. You can attach a lot more than just one or two arrays to the drum. The number of arrays N used on the drum ( 13 ) must be adapted to the respective problem. The detector arrays ( 18, 19, 20, 21 ) are positioned around the axis of rotation ( 14 ) by left ( 16 ) or right ( 17 ) rotation. If the individual arrays are staggered on the drum, the individual images can be averaged to reduce noise. If, on the other hand, single images are used, the respective arrays being offset such that the space between two array sensors is detected, resolution improvements can be achieved. This multiarray arrangement can be used particularly advantageously in connection with parallel computers if a computer is connected behind each array, since existing software and hardware can be used.

Instead of attaching one or more arrays to a drum, a turntable or holder for arrays is also possible. Fig. 7 shows this situation. The axis of rotation ( 24 ) passes exactly through the geometric center of the detector array receiving surface ( 23 ) of the surface sensor ( 27 ). If the individual images are added at the recording points ( 25 ), an averaging is obtained via the individual individual sensors of the array.

This principle is shown in detail in FIG. 8. By rotating the sensor ( 27 ) about an axis of rotation perpendicularly through the receiving surface, a recording is made here after rotation of 90 ° (or 180 °). It is thereby achieved that the individual sensors of the array which are furthest away from the read point are averaged with those close to the read point. This adjustment to a background noise level that is as homogeneous as possible leads to a general improvement in the reading.

FIG. 9 shows how a deflection of a light beam by small angles can be achieved by refraction, that is to say optically. The basis is called the deflection prism, breaking wedge or breaking prism. It consists of a glass body, which is delimited by two flat surfaces intersecting at the prism angle χ . The angle between the entrance and exit beam is called the deflection angle δ . He calculates himself

δ = ε ₁ + ε ₂ - χ

With

If the angle of incidence and the prism are small, δ = χ (n -1) applies with good approximation. Since only small changes in location are required for the purpose used, the errors that occur can be kept within limits.

An embodiment of the variable wedge is shown in FIG. 10, the Abat wedge. Here a plano-convex lens and a plano-concave lens of the same radius made of the same glass are used, which, when placed centrally, form a flat plate without deflection, but pushed apart in the spherical surface but represent a prism. The effective angle α is given by α = v / r ; if both spherical surfaces are separated by d , then d (n -1) must be their difference in radii; it applies

with the focal length f of the lens and the displacement v . In the case of the oscillating wedge pair in FIG. 11, the dependence on the angle of incidence is used to change the deflection, two identical wedges being rotated about a parallel to the breaking edges.

The rotating wedge pair (Herrschel double prism, wedge compensator or diasporameter) in Fig. 12, on the other hand, is based on the following principle: If two identical wedges are turned in opposite directions around a common axis in the main section, the effect of a variable wedge is obtained. If one calculates the twist ϕ from the position of the largest deflection, then for the respective deflection δ = 2 δ _K × cos ρ with the deflection of the single wedge δ _K. The dependence of the angle of rotation is practically linear in a medium range (90 ° ± 45 °), the angle of rotation being a multiple (15 to 20 times) of the deflection angle. Fig. 13 shows a possible embodiment. Two identical prisms ( 29 ) are arranged one behind the other in the beam path ( 30 ), which are rotated in the opposite direction by a side-mounted pinion ( 32 ). In the position shown, the deviation of the incident beam ( 28 ) from the exit beam ( 31 ) is zero; when the beam is rotated by an overall angle of 2 gegens , the beam is deflected δ = 2 α (n -1) sin ϑ .

α - "breaking angle"
n - refractive index.

The pair of rotary wedges is attached between the lens ( 33 ) and the array ( 34 ) with its image-sensitive surface ( 35 ).

Claims (13)

1. A method for data acquisition by means of detector arrays, in which an object is mapped in whole or in part on a detector array and the signals generated by it are further processed, characterized in that, for signal improvement, the object is moved successively via preselectable amounts in the respective detection level via a detector array , whereby multiple measurements are made for an averaging. 2. Method for data acquisition, in which an object is whole or partially mapped onto a detector array and that of this generated signals are processed, thereby characterized in that at least one detector array in succession through rotary movements in discrete recording positions is brought, with a in these preselected positions Signal acquisition takes place. 3. Method for data acquisition, in which an object is whole or partially mapped onto a detector array and that of this generated signals are processed, thereby characterized that the object for signal improvement by means of a between the detector array and the object adjustable optical device for distraction of the rays coming from the object one after the other preselectable amounts by small angles via the detector array is moved, taking multiple measurements for an averaging be made. 4. The method according to claim 2, characterized in that at least one detector array parallel to the optical one Axis extending axis is rotated. 5. The method according to claim 2, characterized in that at least one detector array perpendicular to the optical axis is rotated. 6. The method according to claim 1, 2 or 3, characterized in that  that a detector array regularly between two consecutive Recordings to identify the array characteristic values evenly irradiated (white keying) and / or read out without external radiation (black keying) becomes. 7. The method according to claim 6, characterized in that through signal processing of the regularly determined values and possibly necessary signal delay an output value correction of a detector array. 8. Device for carrying out the method according to claim 1, 2 or 3 with a CCD sensor and optics as well an object lighting device, characterized in that that between the CCD sensor and the object there is a movable filter with several sectors. 9. Device for carrying out the method according to claim 4 or 5 with a CCD sensor and a recording optics as well an object lighting device, characterized in that that multiple shots from preselected positions and the CCD sensor can be rotated on a bracket is appropriate. 10. Apparatus for carrying out the method according to claim 3 with a CCD sensor and a recording optics as well as one Object lighting device, characterized in that the adjustable between the detector array and the object optical device to distract from the object is a rotating wedge. 11. The device according to claim 9, characterized in that several CCD sensors with filters on a drum are attached. 12. The apparatus according to claim 9, characterized in that a rotatable drum is provided which has at least one CCD sensor carries, and that a mechanism for axial This drum floats.   13. An apparatus for performing the method according to claim 1, 2 or 3, characterized in that before the discrete recording position a suitable mass for producing color images is attached.