DE2437529B2

DE2437529B2 -

Info

Publication number: DE2437529B2
Application number: DE2437529A
Authority: DE
Inventors: Guenter Dr.-Ing. 2000 Hamburg Kowalski
Original assignee: Philips Patentverwaltung GmbH
Current assignee: Philips Intellectual Property and Standards GmbH
Priority date: 1974-08-03
Filing date: 1974-08-03
Publication date: 1978-10-26
Also published as: FR2280919A1; DE2437529C3; GB1523764A; FR2280919B1; JPS5138892A; DE2437529A1; JPS5426477B2

Description

F i g. 5 ein Netzwerk zur schnellen elektronischen Drehung des Koordinatensystems.F i g. 5 a network for rapid electronic rotation of the coordinate system.

Das Objekt O werden durch N Röntgenblitze (F\ ... Fn) aus verschiedenen Richtungen durchleuchtet, die durch die zugehörigen Horizontalwinkel (φι ... φκ) und Polarwinkel (ψι ... ψ/ν) definiert seien (Fig. 1). Der Bildverstärker wird so nachgeführt, daß eine Verbindungslinie vom Mittelpunkt des Bildverstärkers zur Strahlenquelle (Zentralstrahl) stets durch einen Punkt (Q) geht. Die Positionen des Bildverstärkers sind in Fig. 1 durch seinen Mittelpunkt (M\ ... Mn) gekennzeichnet. Nach elektronischer Abtastung durch eine Fernsehkamera 2 mit einem optischen System 3 und der Zwischenspeicherung durch einen Bildspeicher 4 werden die Bilder in einer Superpositionseinheit, z. B. einer Speicherröhre 5, mit einem Elektronen-Schreibstrahl derartig entzerrt superponiert, daß nur eine Objektschicht scharf abgebildet wird. Die Entzerrung geschieht dadurch, daß zu den Ablenkspannungen (u_x, u_y) des Ablenkungsspannungsgenerators 6 zusätzliche Korrekturspannungen (Au_x, Au_y) mittels Operationsverstärkern 6', 6" analog addiert werden. Die Korrekturspannungen werden in einem Entzerrungsnetzwerk 7 erzeugt. Dieses Entzerrungsnetzwerk wird durch einen Prozeßrechner 8 programmiert. Die notwenigen Treiberverstärker zur Ansteuerung magnetischer Ablenkeinheiten sind im Schaltbild nicht gesondert angegeben. Die Entzerrung wird an einem Punkt (P) einer beliebigen, zu rekonstruierenden Objektschicht dargestellt. Zu diesem Punkt (P) gehören jeweils für die einzelnen Projektionsbilder (P\ ... P'n) die Spannungspaare der Ablenkspannungen (u_x\... u_xn und u_y\... u_yN) der Fernsehkamera. Die Bildpunkte (P\ ... P'n), die einem Objektpunkt (P) entsprechen, sollen für alle superponierten Bilder koizident auf einen Punkt abgebildet werden, der den Ablenkspannungen U_x und ü_y der Speicherröhre 5 entsprechen soll. Die erforderlichen Korrekturspannungen {Δυ_χη = U_x- u_x„ und Au_y„ = ü_y—u_yn, /7=1 ... N) hier dargestellt in Abhängigkeit von einem Objektpunkt (P), lassen sich nach mathematischer Elimination der Koordinaten dieses Punktes darstellen als Funktionen der Ablenkspannungen selbst, d. h.The object O is illuminated by N X-ray flashes (F \ ... Fn) from different directions, which are defined by the associated horizontal angles (φι ... φκ) and polar angles (ψι ... ψ / ν) (Fig. 1) . The image intensifier is adjusted in such a way that a connecting line from the center point of the image intensifier to the radiation source (central ray) always goes through a point (Q). The positions of the image intensifier are identified in FIG. 1 by its center point (M \ ... Mn) . After electronic scanning by a television camera 2 with an optical system 3 and the intermediate storage by an image memory 4, the images are in a superposition unit, e.g. B. a storage tube 5, superimposed so equalized with an electron write beam that only one object layer is sharply imaged. The equalization takes place in that additional correction voltages (Au _x , Au _y ) are analogously added to the deflection voltages (u _x , u _y ) of the deflection voltage generator 6 by means of operational amplifiers 6 ', 6 ". The correction voltages are generated in an equalization network 7. This equalization network is programmed by a process computer 8. The necessary driver amplifiers for controlling magnetic deflection units are not specified separately in the circuit diagram. The rectification is shown at a point (P) of any object layer to be reconstructed. This point (P) belongs to the individual projection images (P \ ... P'n) the voltage pairs of the deflection voltages (u _x \ ... u _x n and u _y \ ... u _y N) of the television camera. The pixels (P \ ... P'n) , which correspond to an object point (P) , are to be mapped coincidentally for all superposed images on a point which is to correspond to the deflection voltages U _x and _{U y of the storage tube 5. Di.} e necessary correction voltages {Δυ _χη = U _x - u _x " and Au _y " = u _y _{- u yn} , / 7 = 1 ... N) shown here as a function of an object point (P), can be calculated after mathematical elimination of the Represent coordinates of this point as functions of the deflection voltages themselves, ie

Au_xn = ü_x - u_xn = X_n (u_x, u_y) Au_yn = ü_y-u_yn = Y„(u_nu_y) Au _xn = ü _x - u _xn = X _n (u _x , u _y ) Au _yn = ü _y -u _yn = Y "(u _n u _y )

Die Funktionen X\ ... Xn und Y\ ... Yn sind nichtlineare Funktionen der Ablenkspannungen. Um den Fehlereinfluß einer schaltungstechnischen Approximation dieser Funktionen klein zu halten, sollten sie selbst klein sein gegenüber den eigentlichen Ablenkspannungen. Dazu sind zwei Maßnahmen erforderlich, a) Die Wahl eines geeigneten Abbildungspunktes (U_x, ü_y). Dieser Abbildungspunkt sollte möglichst zentral in dem Gebiet liegen, das durch die Ablenkspannungen der einzelnen Bilder (u_x\... u_xn, u_y\ ... u_yn) umschrieben wird. Als sinnvoller Abbildungspunkt kann der Punkt gewählt werden, auf den der Objektpunkt (P) bei einer axialen Zentralprojektion abgebildet würde. Es bleibt dann im synthetisierten Bild der Objektschicht eine Gesamtverzerrung bestehen, die dieser axialen Zentralprojektion entspricht einschließlich eventueller Verzerrungen des Bildverstärker-Kamera-Systems. Im allgemeinen wird diese Gesamtverzerrung ohne Bedeutung sein, sie kann jedoch auch durch fest eingestellte Entzerrungsnetzwerke am Kameraablcnksystem beseitigt werden. (Eine andere Möglichkeit ist der Mittelwert der Spannungen aller einzelnen Projektionen.)The functions X \ ... Xn and Y \ ... Yn are non-linear functions of the deflection voltages. In order to keep the error influence of a circuit-related approximation of these functions small, they should themselves be small compared to the actual deflection voltages. Two measures are required for this: a) The choice of a suitable mapping point (U _x , ü _y ). This imaging point should be as central as possible in the area that is circumscribed by the deflection voltages of the individual images (u _x \ ... u _x n, u _y \ ... u _y n). The point on which the object point (P) would be mapped in an axial central projection can be selected as a sensible mapping point. An overall distortion then remains in the synthesized image of the object layer, which corresponds to this axial central projection, including possible distortions of the image intensifier-camera system. In general, this overall distortion will be of no importance, but it can also be eliminated by means of fixed equalization networks on the camera deflection system. (Another possibility is the mean value of the voltages of all individual projections.)

b) Die Wahl der gewünschten Objektschicht. Günstige Ergebnisse, d. h. kleine Korrekturspannungen ergeben sich z. B. durch Verschieben der Primärschicht K\ (F i g. 2) oder bei kugelkalottenförmiger Primärschicht durch konzentrische Kugelflächenabschnitte K₂(F i g. 2).
Unter Beachtung der Punkte a) und b) lassen sich dieb) The choice of the desired object layer. Favorable results, ie small correction voltages result, for. B. by moving the primary layer K \ (Fig. 2) or with a spherical cap-shaped primary layer by concentric spherical surface sections K ₂ ( Fig. 2).
Taking into account points a) and b), the

ίο Korrekturspannungen (für einen Horizontalwinkel φ — 0) hinreichend genau durch einfache Funktionen annähern, wie z. B.ίο Correction voltages (for a horizontal angle φ - 0) approximate with sufficient accuracy using simple functions, such as B.

X_n = a„ + b_n (u_x - c„f + d„u/ (u_x - e„) X _n = a "+ b _n (u _x - c" f + d "u / (u _x - e")
Y_n = fnu_x ■ u_y Y _n = fnu _x ■ u _y

Die elektronische Erzeugung dieser Korrekturspannungen zeigt Fig.4. Hierbei seien DA\, DA₂, DAj Digital-Analog-Wandler, über die das Netzwerk programmiert wird; Si, S₂, S3 analoge Addierer (z. B. Addierverstärker); M\, M₂, Mi, Mi, sind analoge Multipliziereinheiten und Λ/5, Λ4, Μη sind multiplizierende Digital-Analog-Wandler. Diese können jedoch auch durch jeweils einen Digital-Analog-Wandler in Verbindung mit einem analogen Multiplizierer ersetzt werden. Die Parameter, mit denen das Netzwerk programmiert wird (a„ ... f„) werden durch den Prozeßrechner 8 vorgegeben. Dort können sie gespeichert sein oder auch jeweils neu berechnet werden. Die einzelnen Parameter sind abhängig von den geometrischen Abbildungsgesetzen einer schiefen Zentralprojektion, von der Form der strahlenempfindlichen Schicht, von der Verzerrung des Bildverstärker-Kamera-Systems, von der geforderten Gesamtverzerrung und von der abzubildenden Objekt ■ schicht. Da jeder Parameter von den angeführten Umständen bzw. Faktoren abhängt, müssen alle neu berechnet werden, falls sich einer ändert The electronic generation of these correction voltages is shown in Fig. 4. Here DA \, DA ₂ , DAj are digital-to-analog converters that are used to program the network; Si, S ₂ , S3 analog adders (e.g. adding amplifiers); M \, M ₂ , Mi, Mi, are analog multiplying units and Λ / 5, Λ4, Μη are multiplying digital-to-analog converters. However, these can also be replaced by a digital-to-analog converter in conjunction with an analog multiplier. The parameters with which the network is programmed (a "... f") are specified by the process computer 8. They can be saved there or they can be recalculated each time. The individual parameters depend on the geometric imaging laws of an oblique central projection, the shape of the radiation-sensitive layer, the distortion of the image intensifier-camera system, the required overall distortion and the object layer to be imaged. Since each parameter depends on the stated circumstances or factors, they must all be recalculated if one changes

Die Drehung des Zentralstrahles (F_nQM_n, π = \ ...N) um einen Horizontalwinkel (φι ... φ/ν) wird durch elektronische Drehung des Koordinatensystems berücksichtigt (Fig.3). Das Koordinatensystem der Ablenkspannungen wird zuerst elektronisch gedreht um den Horizontalwinkel (F i g. 3, k) anschließend entzerrt wie für den Horizontalwinkel φ = 0 (F i g. 3, I) und danach wieder um den Horizontalwinkel zurückgedreht (Fig.3, m) Diese Drehung kann z.B. durch Einheiten geschehen, die in der Analogrechentechnik als Resolver bekannt sind. Für schnelle Verarbeitung der Bilder (Fernsehnorm) eignen sich Schaltungen nach Fig.5. Hier sind Ms, M%, M]₀, Mn multiplizierende Digital-Analogwandler und Sa und S5 analoge Addierer. Über dieThe rotation of the central ray (F _n QM _n , π = \ ... N) by a horizontal angle (φι ... φ / ν) is taken into account by the electronic rotation of the coordinate system (Fig. 3). The coordinate system of the deflection voltages is first rotated electronically by the horizontal angle (Fig. 3, k), then rectified as for the horizontal angle φ = 0 (Fig. 3, I) and then rotated back again by the horizontal angle (Fig. 3, m) This rotation can be done, for example, by units that are known as resolvers in analog computing. Circuits according to Fig. 5 are suitable for fast processing of the images (television standard). Here, Ms, M%, M] ₀ , Mn are multiplying digital-to-analog converters, and Sa and S5 are analog adders. About the

^w digitalen Eingänge S\, S₂ und c werden die trigonometrischen Funktionen des Drehwinkels durch den Prozeßrechner vorgegeben, entsprechend s\ = sin φ, S₂ = —S\, c = cos φ. Bei der Rückdrehung sind die Eingänge S\ und » zu vertauschen. Bei festen diskreten Werten der ^With digital inputs S \, S ₂ and c , the trigonometric functions of the angle of rotation are specified by the process computer, corresponding to s \ = sin φ, S ₂ = —S \, c = cos φ. When turning back, the inputs S \ and »must be swapped. With fixed discrete values of the

^r>^r> Horizontalwinkel (φι... φ/ν) können die multiplizierenden Digital-Analog-Wandler durch Widerstandskaskaden mit festen Abgriffen ersetzt werden. Der Prozeßrechner steuert dann die Abgriffe der Kaskaden.
Bei einer kreisförmigen Verwischungsfigur (d. h. alle ^r > ^r > horizontal angle (φι ... φ / ν), the multiplying digital-to-analog converters can be replaced by resistor cascades with fixed taps. The process computer then controls the tapping of the cascades.
In the case of a circular blurring figure (i.e. all

M> Positionen der Strahlungsquelle Fi... F/v[F i g. 1] liegen auf einem Kreis) brauchen bei der Superposition für jedes einzelne Projektionsbild nur die Parameter des Horizontalwinkels bei der elektronischen Drehung des Koordinatensystems neu gesetzt werden. Die anderenM> Positions of the radiation source Fi ... F / v [F i g. 1] lie on a circle) only need the parameters of the superposition for each individual projection image The horizontal angle can be reset when the coordinate system is rotated electronically. The others

^Μ Parameter (a ... f) werden nicht verändert, wenn der Schirm als auch die abzubildende Objektschicht rotationsymmetrisch ist. ^Μ Parameters (a ... f) are not changed if the screen and the object layer to be imaged are rotationally symmetrical.

Hierzu 5 Blatt ZeichnungenIn addition 5 sheets of drawings

Claims

Patent claims:

1. Process for the synthesis of any layers of a three-dimensional object by superposition of images, which by central projection of the object from different directions onto a Arbitrarily curved radiation-sensitive layer produced, electronically scanned and in one Memory are stored, characterized in that the individual projection images are then superimposed in one Medium, e.g. B. a storage tube, electronically equalized so that all points are free selectable object layer are mapped coincidentally, with a programmable, by a Process computer controlled equalization network when superimposed on the deflection voltages of the electronic sampling with the help of two analog adding amplifiers each for the two orthogonal Additional correction voltages added to the deflection directions from the desired object slice depend, and these sums of the original deflection voltages with the associated Correction voltages used to deflect a write beam for the overlay medium will.

2. The method according to claim 1, characterized in that object layers are imaged, which in the borderline case continuously merge into the primary layer, whereby the layer counts as the primary layer, the points of which are mapped coincidentally during the superposition without correction voltages and the is a layer similar to the radiation-sensitive layer.

3. The method according to claim 2, characterized in that object layers are imaged, which are determined by the displacement of the primary layer in the axial direction.

4. The method according to claim 2, characterized in that that in the case of a spherical cap-shaped radiation-sensitive layer, object layers are imaged which are concentric spherical surfaces and have the same center as the primary layer.

5. The method according to claim 1, characterized in that for opposite to the deflection voltages small correction voltages the individual pixels of a desired object layer coincide are superimposed with such corrected deflection voltages, which lie in the range that is determined by the Deflection voltages is limited when reading out the various projection images the individual Object points were assigned.

6. The method according to claim 5, characterized in that the pixels are corrected with such Deflection voltages are superimposed from the mean values of all the individual object points deflection voltages associated with the various projection images are formed. t> o

7. The method according to claim 5, characterized in that the pixels are corrected with such Deflection voltages are superimposed, assigned to them in an axial central projection would be.

8. The method according to claim 5, characterized in that the pixels are corrected with such Deflection voltages are superimposed by the mean values from the deflection voltages of the actual oblique central projection and the deflection voltages of the oblique central projection can be defined from the opposite direction.

9. The method according to claim 1, characterized in that in a rotationally symmetrical radiation-sensitive layer the rotation of the central beam around an axis of rotation with a Horizontal angle due to the electrical rotation of the coordinate system of the Abienkspannungen around the equal horizontal angle, a remaining equalization and subsequent reverse rotation and the remainder of the equalization network works without parameters that depend on the rotation depend on this horizontal angle.

10. The method according to claim 9, characterized in that for the correction of the deflection voltage, which corresponds to a direction perpendicular to the plane of the central ray axis of rotation ([χ direction), a voltage is chosen which is the weighted product of the two deflection voltages corresponds, and that for the correction of the orthogonal deflection voltage (^ direction) a voltage is chosen which corresponds to the sum of the weighted square of the deflection voltage of the .v direction plus a constant voltage and the weighted square of the deflection voltage the y-direction multiplied by the voltage of the x-direction plus a constant voltage and an additional constant voltage.

11. The method according to claim 10, characterized in that in the equalization network further higher products or mixed products of the deflection voltages generated and these for correction to be added.

12. The method according to claim 1 or the following, characterized in that the analog Operations are performed with currents instead of voltages.

13. Arrangement for performing the method according to claim 9, characterized in that for Components and digital-to-analog converters with subsequent analog multipliers for the weighting factors or multiplying digital-to-analog converters are used, while the constant Voltages are specified via digital-to-analog converters with subsequent analog summation.

14. Arrangement for performing the method according to claim 9, characterized in that for electronic rotation and reverse rotation of the coordinate system by four rapid multiplying Digital-to-analog converter or four digital-to-analog converters with subsequent analog multipliers are provided that the actual deflection voltages with the trigonometric Multiply the functions ± sin, cos of the angle of rotation and then add them analogously Products of the deflection voltages generated for a rotated coordinate system, according to the math formulas

U _x . rotated = U _x COS ψ - Uy Sin φ, Uy, rotated = UySing) + ü> .COS <p,

whereby the trigonometric values of the angle of rotation are given digitally by the process computer are.

15. The arrangement according to claim 14, characterized in that the fixed angles of rotation

multiplying digital-to-analog converter through resistance-voltage divider with fixed taps are replaced.

The invention relates to a method and an arrangement for the synthesis of any layers of a three-dimensional object by superposition of images which are generated, electronically scanned and stored in a memory, as a central projection of the object from different directions onto an arbitrarily curved radiation-sensitive layer.

It is known that by projecting a three-dimensional object from different directions on a planar radiation-sensitive layer the image of a planar and to the radiation-sensitive layer Reconstruct the parallel object layer by shifting the superimposition of the individual projection images can. (Lit .: D. G. G r a η t, Tomosynthesis: A Three-dimensional Radiographic Imaging Technique, IEEE, BME-19,1 [1972], 20-28 and E. R. Miller, E. M. McCurry, An infinite Number of Laminograms from a finite Number of Radiographs, Radiology, 98, Feb. 1971, 249-255 and K. Dümmling, A new method for multiple layers with the help of TV image storage, Der Radiologe 9, 37 [1969].) The first object layer is created by not shifting Overlay of all individual images (primary layer). For a curved screen you could do without one Loss of resolving power only this primary layer can be reconstructed (Lit .: H. Vie ten, Über the dependence of the depicted object layer on the shape of the radiation-sensitive image layer, Stratigrafia, 2 [1957], 2-8).

The method of reconstructing object layers by shifting overlay has the disadvantage that one has to rely on a flat screen or that one has a considerable deterioration in the resolution receives. Radiation-sensitive layers with a flat surface have too low a resolution, poor gradation (MicroChannel electron multiplier) or are too insensitive (image converter). To build an electronic tomosynthesis method must therefore be used an image intensifier usually has a curved radiation-sensitive layer.

The object of the invention is to provide a tomosynthesis with a high resolution even when using to achieve powerful and highly sensitive image recording devices.

This object is achieved in that the individual projection images are electronically scanned in one Memory stored and then with an overlay in a medium, z. B. a storage tube, electronically rectified so that all points of a freely selectable object layer are mapped coincidentally with a programmable equalization network controlled by a process computer the superposition of the deflection voltages of the electronic scanning with the help of two analog Adding amplifier each with additional correction voltages for the two orthogonal deflection directions which depend on the desired object layer, and these sums of the original Deflection voltages with the associated correction voltages for deflecting a write beam for the overlay medium can be used.

The images of the individual projections are rectified in such a way that each object point has one desired slice in the superposition of the individual images always mapped in the same image point and each projected image is electronically scanned before being superimposed by a programmable equalization using the deflection voltages, e.g. B. a storage tube, electronically equalized.

The invention has the following advantages.

a) A quick recording sequence can be achieved because the images of the individual projections are initially stored distorted.

b) It can be done with the help of the specified procedure by appropriate programming of the equalization network any object layer can be reconstructed (e.g. also a plane layer). The choice of the object layer will not Restricted from the start, as if the distortion was corrected during the recording of the projection images.

c) Are the object slices to be reconstructed suitably selected (e.g. by shifting the

Primary layer), the necessary non-linear corrections to the deflection voltages are made small compared to these themselves and the equalization network can be constructed correspondingly simply .in be.

d) If a rotationally symmetrical screen is assumed, there is a further advantage. By electrical rotation of the coordinate system of the deflection voltages can be the parameters that

S5 are required for programming the equalization network, separate into those that only depend on the object layer to be reconstructed and a parameter that describes the rotation of the tomography device by a horizontal angle • to (ψ) . In the case of a circular blurring figure, only the parameter that describes this angle needs a; in front of the superposition during a synthesis; Projection image can be reset.

-ts e) Through the network can be without additional effort also correct distortion errors in the image intensifier camera system. If a rotationally symmetrical Screen is used and the rectification after the electronic rotation is done so the coordinate system can, however also only rotationally symmetrical distortions can be corrected.

f) The achievable resolution can be achieved through a correspondingly complex equalization network in any case can be increased to such an extent that the overall resolution of a real system is achieved the influence of errors on other components is limited.

The invention is explained with reference to the figures. It shows

F i g. 1 a principle sketch of the procedure,
F i g. 2 two examples of object layers that lead to small corrections of the deflection voltages,

F i °. 3 is a block diagram of the entire Entzerrungsb ^r> network

F i g. 4 the rectification network, but without the required rotation and reverse rotation of the coordinate system and