DE3829300A1 - Method and device for scanning an object - Google Patents

Abstract

The invention relates to a method and a device for scanning the terrain surface and other objects transversely to the direction of flying or movement of the scanner with a detector row arranged at right angles to the scanning direction, which row contains a number of individual detectors of the same size, the terrain or object strip to be scanned being subdivided into different distance ranges within which the elementary detector signals are in each case combined both within the detector row and within successive exposure periods by averaging in such a manner that the combined individual signals in each case form or cover an object pixel of approximately constant size independent of the scanning range, these object pixels always being normal (perpendicular) to the scanning beam and each scan generating a terrain or object strip with approximately equal object pixel size and constant object pixel number in width and length of the scanning strip. This method and this device eliminates the known defect of opto-mechanical scanners, the scanning strips of which exhibit a variable strip width in dependence on the scanning range. <IMAGE>

Description

The invention relates to a method and a Ground surface scanning device and other objects transverse to the direction of flight or movement with one arranged transversely to the scanning direction Row of detectors, several, of equal size Contains individual detectors. This need arises especially for military reconnaissance and for the Remote sensing in civilian areas. For this you use u. a. so-called mechanical-optical scanner, their Working principle is generally known (see e.g. Hofmann, O .: Image quality of active and passive scanners, Image measurement and aerial photography 51, 1983, Issue 3, pp. 103-117). One or more detectors in are arranged in the image plane of a lens the terrain across the flight direction line by line. The Scanning takes place in front of the lens arranged optical deflection mechanism, such as a Rotating mirror. By the forward movement of the wearer lined up row by row transversely to the flight direction, so that a terrain or object strip is scanned.

A major problem with this process, especially when flying low or at large angles, consists of that the distance from Scanning to the surface of the site or to the object strongly changed and consequently also that of the detectors recorded object point size, provided focal length and detector remain constant. This creates the well-known Coverages with a growing angle of view, since those of terrain strip scanned with a detector growing angle of view become ever wider.  

In principle, constant resolution of the object can changing object distance by changing the Focal length or the detector size can be achieved.

A well-known and from the company Texas Instruments realized solution (see J.-M. Kaltenbach "special Procedure to avoid geometry errors in IRLS ", Carl Cranz Society, 47th year p. 1.08, Nov. 1986) consists of individual detector Assemble part detectors and according to the Remove the signals from the individual sections retrieve. It is a special one Detector, the sections of which have different shapes and Have sizes and only in one scan allows the scanning of a line in the scanning direction.

The invention has for its object a method and a device of the type in question specify with which over the entire scanning range constant resolution is achieved, the overflown Object can be scanned seamlessly and that Overall system flexible to different parameters, e.g. B. the flight altitude, can be adjusted.

This object is achieved according to the invention by characterizing features of claims 1 to 5 solved.

Accordingly, the signals from individual detectors become like this summarized that over the entire scanning range Object pixels approximately constant from that Independent-sized scanning distance can be formed.  

The given solution allows with a number equal large detectors simultaneously scan several, parallel scan lines at approximately distance-independent object point size (pixel size). Simultaneous scanning across the flight direction lying parallel tracks is especially then required if large at high airspeed geometric resolution, d. i.e., small object pixel sizes, be requested.

In principle, the process consists of a so-called "Resampling" a pixel grid, i. that is, the conversion an uneven pixel grid into another, regular.

At the largest scanning distance E _max , the so-called "elementary object pixel size D " corresponds to the required "constant standard object pixel size B ". Proportionally as the distance E decreases, the elementary pixel size D becomes smaller, and the standard pixel size B is formed in each case by optimally combining (resampling) elementary pixels D.

This summary, ie the averaging of a plurality of elementary pixels D to form a standard pixel B , at the same time improves the signal-to-noise ratio of the object standard pixel B.

At the maximum distance E _max , as already explained, an elementary pixel D corresponds to a standard pixel B , an improvement in the signal / noise ratio is therefore not possible within one scan.

Now, especially at the maximum distance E _max, the signal / noise ratio is the most critical and there is a need to improve it there too.

According to a preferred embodiment of the invention this goal is achieved in that with increasing overlap overlap neighboring scanning strips is used and the Multi-sampled terrain pixel signals summarized, d. that is, integrated or averaged.

Further refinements of the invention result from the Sub-claims emerge.

The invention is in exemplary embodiments based on the Drawing explained in more detail. In this represent:

Fig. 1 is a schematic representation of a conventional scanning a terrain surface from an aircraft;

Figure 2 is a schematic illustration of a strip of land with conventional or inventive sample.

Fig. 3 is a schematic representation of a scan of a terrain strip transverse to the direction of flight of an aircraft for explaining important for the invention system parameters;

Figure 4 shows a strip of the object pixels generated by the scan;

Figure 5 is a schematic representation of the scanning at which the creation of equally sized object pixel is explained.

FIG. 6 shows a representation similar to that in FIG. 3 to explain the generation of object pixels of the same size; FIG.

Fig. 7 is a block diagram of a first embodiment of a device according to the invention for scanning a terrain surface;

Figure 8 is a schematic representation of a scanning region for explaining a scanning method with improved signal / noise ratio.

9 is a block diagram of a second embodiment of a device according to the invention for scanning a terrain surface with improved signal / noise ratio.

FIG. 10 shows a block diagram of an integration processor used in the device according to FIG. 9.

In a conventional scanning of an object, e.g. B. an off-road surface from an aircraft, the scanning areas recorded by a scanner show a shape labeled AB in FIG. 2. The scanning areas extend perpendicular to the X axis, which is parallel to the flight direction, in the Y direction and broaden with increasing distance from the X axis. The area areas recorded by the individual detectors of a scanner arranged in the aircraft thus vary in size depending on the respective distance between the scanner and the viewed area area. The output signals of the individual detectors are processed in accordance with the invention in such a way that the scanning area, as indicated by hatching, becomes a rectangle with the width B _n and constant resolution is achieved within this scanning area, ie the scanning area is divided into object pixels each having the same pixel size.

It is thus required that the object pixel sizes, the are each normal to the scanning beam, approximately are independent of distance.

In the image plane of the scanner, a number of identically sized detectors are arranged in a row at right angles to the scanning direction. The scanning beam moves in a plane that is normal to the direction of flight and is approximately identical to the plane of the drawing in FIG. 3.

The size of the individual detectors and the focal length of the scanner are usually dimensioned such that the required object pixel size is covered by a detector and an exposure period at the greatest object distance (transverse to the direction of flight). The detector width d determines the "elementary object pixel size" D _x in the flight direction, the elementary object pixel height D _y in the scanning direction is determined by the exposure period Δ t . As a rule, D _x and D _y are chosen to be the same size. There are simple relationships:

D _y = · E · Δ t (2)

Mean in it

E = object point distance
F = focal length
d = detector width
= Scanning angular velocity
Δ t = exposure period (exposure interval of an elementary detector signal

The aim of the invention is that during a scan n parallel, equally large and always constant number of object pixels are generated and stored in an image memory, that is to say that a scan forms an image strip lying at right angles to the flight direction. The width B _{n of} this strip (in the object) is

B _n = n B _x (3)

where B _{x represents} the desired width of the object pixel.

In the scanning direction Y , too, approximately constant object pixel sizes B _{y are to be} formed, irrespective of the instantaneous image angle, that is to say of the scanning distance.

The object distance changes continuously during a scan, and the object pixel size according to (1) and (2) shrinks as the distance E becomes shorter.

In order to compensate for this known effect, the entire distance range is now subdivided into a number of sub-areas, within which signals from a number of adjacent detectors and a number of exposure periods are combined by averaging in such a way that the combined elementary detector signals or exposure periods each optimally cover an area that corresponds to the specified one Correspond to object pixel size B _x and B _y .

With the help of FIGS. 5 and 6 this is made clear by means of examples. It should be noted that these images are very distorted. The distances E in the Y direction are shown very shortened in relation to the pixel dimensions D and B, respectively.

In Fig. 5, the line of symmetry of the scanning track is the Y axis. The constant object pixel size is B _x . 4 ground pixels I, II, III, IV are formed on the left (L) and right (R) of this Y axis. The detectors left (L) and right (R) are 1, 2, 3, 4. . . referred to, the distance areas A , B , C,. . .

Table 1 shows which detector signals are combined within a range in order to generate the ground pixel width B _x .

Table 1

So z. B. in the distance range E, the object pixels B _{x of} the row IV are generated by combining the elementary pixels D _{x of} the detectors 6 and 7 . This area is hatched in FIG. 5.

In principle, the same procedure is followed in the scanning direction, here too the signals of successive elementary exposure periods Δ t or elementary angle increments Δω = · Δ t are combined and averaged, so that object pixels B _{y of} approximately the same size are also formed in the Y direction over the entire scanning range , it should be noted that these object pixels B _{y are} each perpendicular to the corresponding scanning beam. FIG. 6 explains this process, it corresponds to FIG. 3.

The elementary angular increment corresponding to a sampling or clock period Δ t is Δω = · Δ t , the sampling speed representing; Δω is constant for at least one scan. For the greatest scanning distance E _max , an object pixel B _y corresponds to an elementary pixel D _y = E _max · Δω . The shorter the distances, the more angular increments Δω are associated with an object pixel; in FIG. 6, there are, for example, in the drawn distance E three pixel element D _y and three angular increments Δω, which are combined into one object pixel B _y.

Angular ranges can now also be found hereω _A′,l _B'. . . or distance rangesA′,B′ Etc. (inFig. 6 not shown) determine within which a constant number of scanning signals (each correspond to a cycle or angular increment) be summarized and averaged.

These areas can, but need not be, identical to the pixel areas A , B , C. . . in the X direction. Since the distances change continuously during a sampling cycle, but the number of pixels only in full number of pixels, certain inaccuracies arise in the border areas, which always occur more or less during a resampling process and can also be seen analogously to this in FIG. 5.

If, for example, such a scanning device is designed, as already mentioned above, for the largest distance to be scanned or the largest image angle, the elementary object pixel sizes D _x and D _{y are} the desired composite object pixel sizes that are approximately constant over the entire scanning area B _x and B _y correspond, then for the shortest distance E _min to form an object pixel B _x according to equation (1)

Detector signals and according to Eq. (2)

Exposure periods are summarized.

Accordingly, the row of detectors must be at least

r = m · _x n (6)

Contain individual detectors if the object pixels B are to be completely covered by elementary pixels D in all distance ranges.

The range limits A , B , C. . . or A ', B ', C '. . . can, but need not, be identical.

The total number S 'of all object pixels in the scanning direction results from the sum of the object pixel numbers in the individual areas

p′ _A+p′ _B+p′ _C. . . . =S ′      (7)

A scan strip thus contains the constant number of

n · S '(8)

Object pixels (see Fig. 4).

The distance rangesA,B,C. . . . respectively.A′,B′,C.'. . . and partial object pixel sumsp′ _A,p′ _B,p′ _C. . . . are usually different sizes, they depend on Angle of view and the flight altitude.

The overall image strip generated by the scanner sets from the seamlessly consecutive, across Partial strips lying in the direction of flight together. Around this gapless and overlap-free partial stripe sequence the following condition must be met:

v = F _A · B _n (9)

in which
v [m / s] = airspeed
F _A [l / s] = sampling frequency
B _n [m] = partial strip width
represent.

The proposed principle of operation can be started optical-mechanical scanners of the most varied Realize designs. There is only one row of detectors for this with individual detectors of the same size perpendicular to the direction of flight to be arranged in the image plane of the scanning lens (see Eq. 6).

The parameters of the scanner
Focal length F
Detector element size d
Number of detectors r
Number of scan tracks n
Scanning speed (angular speed)
Exposure period Δ t
Distance ranges A , B , C. . . A ′, B ′, C ′ etc.
result from the flight parameters (flight height h , flight speed v) , the image angle transverse to the flight direction and the required object pixel size B.

The distance ranges are the corresponding ones Angle of rotation rangesΔω _A,Δω _B,Δω _C. respectively.Δω _A′,Δω _B′,Δω _C.′ assigned (seeFig. 3). You let yourself be constant rotational speed of the scanner from the Rotation time for these anglest _A,t _B,t _C. . . . respectively.t _A′,t _B′, t _C.'. . . in relation to a reference mark (index) determine. This reference mark can depend on the rollage of the wearer can be variable and each Define plumb line. For more accurate measurements, these can be Angle of rotation of one with the axis of rotation of the scanner coupled angular position transmitter can be tapped.

The detector signal processing and the overall function of the scanner can be seen from FIG. 7.

The analog, elementary detector signals from a plurality of detectors 1 arranged in a line perpendicular to the scanning direction are digitized by an A / D converter 2 and stored in a first memory 3 . Each detector signal is represented by a number, e.g. B. represents a binary number with 8 digits (8 bits) and corresponds to an elementary pixel. A first processor 4, controlled by a program module 8, takes the elementary detector signals of an exposure period from the memory 3 and detects them as a function of the distance ranges A , B , C. . . etc. together to form an object pixel. Each of these composite n pixels corresponds to an object pixel width B _x .

These n pixels are fed to a second memory 5 , which can hold the pixels of several exposure periods. A second processor 9 , also controlled by the program module 8, takes several exposure periods within each of the n pixel sequences depending on the distance ranges A ', B ', C '. . . together and feeds them to a third memory 10 , each of which stores the pixels of a scanning strip. The Y dimension of each composite pixel now corresponds to the object pixel size B _y . The pixels can now be read out of the memory 10 and stored in a mass memory 11 , e.g. B. on a magnetic tape.

The program module 8 is controlled by the angular position of the scanner or by an angular position transmitter 6 connected to it, in order to be able to detect the corresponding distance ranges in each case.

The optimal range and pixel combinations depend, among other things, on the flight height h above ground. The optimal distance ranges and pixel combinations are defined in tables for arbitrarily graded height ranges and these are stored in the program module 8 , which is controlled by an altimeter 7 .

The object pixel resolution B _x , B _{y can also be} changed via this program module 8 , that is to say that different object pixel resolutions can be set or programmed.

This method and device have the following Benefits on:

• 1. The object pixels that are normal to the scanning beam are almost the same regardless of the distance large.  
• 2. A scan stripe can contain one or more parallel rows of pixels n .
• 3. Each scanning strip always contains a constant number of object pixels in the X and Y directions, and the strip width B is constant.
• 4. The scanning strips can therefore be on one the flight speed matched Scanning speed without gaps and Connect to each other without overlap.
• 5. As detectors can be commercially available Detector rows, also integrated detectors (e.g. TDI or sprite detectors) can be used. The Number of individual detectors is practically in the Range from 20 to 200, that depends the respective requirements.
• 6. The process is very flexible. By the corresponding programming can be done Object pixel size changed and almost independent be made from the altitude. This adjustment can be done automatically during the flight.

An extended scanning method is explained with the aid of FIGS . 8 to 10, with which the signal / noise ratio can be improved, in particular for large scanning distances E.

A scanning strip with the width B _n = n · B _x consists of n (in the example n = 2) lines of the standard pixel width B _x , ie

B _n = 2 x B _x (1)

At the minimum distance E _min , M elementary pixels D _x (in the example M = 8) are combined to n = 2 standard pixels B _x . At the maximum distance, D _max = B _x . The scanning width of all M detectors is therefore for E _max

B _M = M * W _x = M * D _max (2)

While for E _min the scanning strips are lined up without gaps and without overlap, the trapezoidal neighboring strips overlap with increasing distance. It is therefore not only the standard pixels b _s , m which lie in the stripe area B _{n that} are formed in a stripe scanning (in FIG. 8 m = 4.5), but as far as possible and requires the signal / noise ratio, from one certain distance, e.g. B. E _with , up to the maximum distance E _max also the standard pixels b _s , m (m = 1 ... M) overlapping on both sides up to a maximum strip width B _M = M · B _x .

In the relationships shown in FIG. 8, the maximum distance E _max results in an r = 4-fold coverage or scanning of each terrain pixel (e.g. G _{s, 7} and G _{s , 8} , where s = f (E _max ) with f as Function characters).

How outFig. 8 can be seen in the removalE _With  the coverage of adjacent strips is less, for this distance can e.g. B. the pixels _jb_{s, 1}, _jb_{s, 2nd}, _jb_{s, 7}, _jb_{s, 8} not be formed and it is only one double integration possible. In this case only the pixels _jb_{s, 5}+ _{j + 1}b_{s, 3} respectively. _jb_{s, 6}+ _{j + 1}b_{s, 4}  added. The reduction factor here is 1 /r= ½, he is depending on distance, the terrain pixelsG _{s, 5} andG _{s, 6} (see= f (E _With)) are only scanned twice.

From theFig. 8 one can see that the pixels _j-1b_{s, m}  one scanj-1 each aroundn shifted pixels _jb_{s, m} overlay the next scan.

To implement this method are therefore the corresponding pixels of adjacent scan strips expedient to store digitally and gradually integrate.

Fig. 10 shows a basic embodiment of an integration processor P ₃ for this integration process, the relationships of Fig. 8 are based. FIG. 10 demonstrates only the working principle.

FIG. 9 shows the block diagram of the overall system, the individual components, insofar as they correspond to those in FIG. 7, being provided with the same reference symbols.

In principle, the following processes take place for the pixel row of a distance step s of a stripe scanning cycle j :

• a) Transfer of the pixels _jb_{s, m} (m= 1 toM) one Row of pixelss of a scanning stripj in a Pixel register21st.
• b) Transfer of the pre-integrated pixels _{j-1 s, m}  (m=n+1 toM) from the integration memory 13 into a memory register22.
• c) Addition of corresponding pixels in the register 21st and22 in an adder23 such that the pixels each _jb_{s, m} of the input register21st  to the pixels _{j-1 s, m + n} of the memory register22  be added so that the new integrated pixel _{j s, m} (m= 1bisM) in one Integration register24th arise.
• d) reading the pixels _{j s, 1} to _{j s, n} from the Integration register24th and multiplication with Using a multiplier25th with the Reduction factor 1 /r, in whichr the number of Integrations(r is distance-dependent) and takeover of the final, reduced pixels _{j , m} (m= 1 to n) into an intermediate register27th.
• e) intermediate storage of the integrated pixels _{j s, m}  (m=n+1 toM) in the integration memory13.

The integration memory 13 is constructed in principle of a matrix with S rows for each distance level s and M - n columns; see. Fig. 9.

In the present example, a fourfold integration takes place, provided that all m pixel spaces are occupied. For example, the signals result

_{j ,1} and_{, 2nd} for the terrain pixels

G _{s, 7} and G _{s, 8} from the following integrations:

G _{s, 7}: _{j ,1}= ¼ (₁b _{s, 7}+ ₂b _{s, 5}+ ₃b _{s, 3}+ ₄b _{s, 1})

G _{s, 8}: _{j , 2nd}= ¼ (₁b _{s, 8}+ ₂b _{s, 6}+ ₃b _{s, 4}+ ₄b _{s, 2nd})

This method can also be modified for a lower degree of integration, which can be done simply by reducing the number of pixels M to be formed.

FIG. 9 shows the principle of operation in connection with the functional sequence shown above for FIG. 7.

For distance ranges that require integration adjacent scanning strips are required in the manner described above via the memory3rd and 5 and the processors4th and9 each for one Removal steps the pixel signals _jb_{s, m} (m= 1 to M) formed and about the store10th in the Input register21st of the integration processor12th  read. At the same time, the Integration memory13 those already pre-integrated pixel _{j-1 s, m} (m=n+1 toM) into the storage register 22 read. After adding, reducing and reading out the final integrated pixels _{j , m} (m= 1 ton) in the cache27th the content of the Integration register24th in the integration memory13  saved.  

The integration memory13 must thereforeS Lines and(M-n)  Have columns, each place an integrated Signal value _{s, m} contains.

Claims (10)

1. Method for scanning an object, e.g. B. a terrain surface transverse to the flight or movement direction of a scanner with a perpendicular to the scanning direction arranged detector row, which contains several equal detectors, characterized in that the terrain or object strip to be scanned is divided into different distance ranges within which the signals of individual detectors (elementary detector signals) are combined both by means of averaging both within the row of detectors and successive exposure periods such that these combined individual signals each form an object pixel of approximately constant size independent of the scanning distance, these object pixels always being perpendicular to the scanning beam and each scan generates a terrain or object strip of approximately the same object pixel size and a constant number of object pixels in the width and length of the scan strip. 2. The method according to claim 1, characterized in that that by combining the detector signals one or more rows equal to one scan large parallel rows of object pixels are formed. 3. The method according to claim 1 or 2, characterized in that also in the coverage areas of adjacent scan strips for each scan strip ( j) a regular pixel grid of the same size standard object pixels (B) is formed and for the purpose of improving the signal-to-noise ratio the pixel signals (b ) the standard pixels (B) of adjacent strips, which cover the same terrain pixels (G) , are integrated. 4. The method according to claim 3, characterized in that

• a) within a scanning strip j standard pixel signals formed _(j b _{s, m;} m = 1 to M) of a removal stage (s) which may maximally contain up to M pixel, from an intermediate register (10) in the input register (21) of a Integration processor ( 12 ) are read in,
• b) the pre-integrated pixels ( _j-1 b _{s, m} ; m = n +1 to M) of the preceding scan (j -1) from an integration memory ( 13 ) into a memory register ( 22 ) of the integration processor ( 12 ) are read in,
• c) the pixels( _jb_{s, m}) of the input register (21st) With the umn Pixels offset pixels( _j-1 b _{s, m};m=n+1 toM) of the memory register (22) about one Adder (23) are added so that the new integrated pixels( _jb_{s, m};m= 1 toM) in one Integration register (24th) of Integration processor (12th) arise
• d) the pixels ( _j b _{s, m} ; m = 1 to n) are read out of the integration register ( 24 ) and the final pixels ( _j b _{s, m} ; m. ) are multiplied by a multiplier ( 25 ) with the reduction factor 1 / r = 1 to n) are generated and stored in an intermediate register ( 27 ), where r represents the number of integrations that are supplied by a control computer ( 26 ) to the multiplier,
• e) the pixels ( _j b _{s, m} ; m = n +1 to M) are taken from the integration register ( 24 ) into the integration memory ( 13 ),

and that for each distance step (s) and for each scan (j) these processes a) to e) are repeated. 5. Device for performing the method according to one of claims 1 to 4, characterized by

• a) a first memory ( 3 ) for storing the signals of the individual detectors ( 1 ) within an exposure period
• b) a first processor ( 4 ) for combining the signals of the individual detectors ( 1 ) in groups;
• c) a second memory ( 5 ) for storing the signals corresponding to the groups of successive exposure periods;
• d) a second processor ( 9 ) for combining individual groups on at least one exposure period for generating object pixels and
• e) a mass storage device ( 11 ) for storing the object pixels.

6. The device according to claim 5, characterized in that a program module ( 8 ) for controlling the processors ( 4 , 9 ) is provided, which is connected to an angular position transmitter ( 86 ) of the scanner and an altimeter ( 7 ), from which the distance and forms height-dependent detector signal combinations and controls the processors ( 4 , 9 , 12 ) based on these signal combinations. 7. The device according to claim 6, characterized in that the angular position transmitter ( 6 ) always provides the scanning angle related to the plumb line between the scanner and the object. 8. Device according to one of claims 5 to 7, characterized in that an integration register ( 13 ) is present in which the integrated pixel values ( _j b _{s, m} ) are stored in a matrix, the memory ( 13 ) (M - n) Columns and S rows (S = number of distance levels). 9. The method and device according to claim 8, characterized in that the multiplication of the integrated and from the integration register ( 13 ) read pixels (b) by a multiplier ( 25 ), the multiplication factor (1 / r) is dependent on the distance by a control computer ( 26 ) is controlled.