GB2268023A - A reconfigurable-format image acquisition system - Google Patents

Abstract

A camera, for use in a machine vision system, comprises a tube imager, means (7, 8) for adjusting the spot size, means (6) for rotating the raster generated by deflection circuitry (5), means (4, 17) controlling the beam so as to switch it on/off, and an image format controller (2) for storing signals to control the spot size, scan position/orientation and beam sampling rate. Switching between formats can be achieved quickly by reading out different sets of control signals stored in the format controller (2). <IMAGE>

Description

A Reconfigurable-Format Image Acquisition System This invention relates to a technique of and an apparatus for acquiring image of different formats: size, sampling pattern, resolution and orientation. This invention is generally used in conjunction with camera tube imagers.

Most machine vision systems make use of cameras and frame-grabbers which are compatible to Television Broadcast Standard. By confining the image scanning format to that of Television Broadcast Standard, one is able to take advantage of the abundant of equipments such as cameras and monitors that are already available in the market.

However, from machine vision application point of view, this has many disadvantages, the most notable limitation is the lack of flexibility. For examples, the TV picture aspect-ratio, that is the ratio between the horizontal and the vertical size of the TV picture, may be pleasant for human viewer but has obvious setback when objects are to be measured. Each picture element or pixel in an image is not square in shape. Problem also arises due to mismatch between the number of picture element in the camera and the frame-grabber which generates interference fringes. Interlacing of lines which reduces TV transmission bandwidth poses some difficulties for machine vision especially in high-speed and precision applications. Instead of capturing a complete picture in 1/50 sec (CCIR) or 1/60 sec.(RS-170), it takes twice as long.However, the most notable limitation is imposed by the standard raster format.

Recent advances have seen the emergent of cameras designed for machine vision application which are programmable to handle different scanning timings other than the standard format. In the field of machine vision, research has found that other image formats in term of non-standard sampling and resolution are more effective in solving some vision problems than others. To take advantage of the non-standard image formats faces difficulties as current image acquisition systems, i.e. camera and frame-grabber, cannot be reconfigured by the user.

Recent research has advocated the strength of the human retina which is essentially circular in shape and has variable resolution changing from the fovea to the peripheral . In other words, the image that is captured by the human retina is a space-variant image.

Customised cameras are built to satisfy such unique image specification. It is obvious that a customised camera is not economical and lacks flexibility. Other than building a customised camera for capturing image of non-standard format, many have attempted in using software and additional hardware to convert the standard format image into a non-standard one, by the so called re-sampling method. The conversion process takes up substantial computational effort and introduces distortions to the image.

There are instances where an image, in standard or non-standard format, needs to be rotated and translated in real-time. Image rotation and translation are very important preprocessings steps in machine vision applications. One of its use is in the registration of images which is crucial to many image analysis tasks. For example, once an image has been registered with respect to a reference image, these two images can then be compared directly. Industrial inspection making use of machine vision often employs direct image comparison or subtraction technique to test whether an object conforms to the reference object. Although image rotation and translation can be performed by software or dedicated hardware, they add cost to the system. Moreover, it is a known fact that software rotation and translation of a discretised image introduces distortion to the resultant image.

The figures associated with the present invention are described in the following: Fig. 1 The overall block diagram of the reconfigurable-format image acquisition system.

Fig. 2 A block diagram of a typical vidicon tube and its associated circuits.

Fig. 3 An illustration of the principle of electromagnetic focusing of an electron beam.

Fig. 4 The schematic block diagram of the image format controller, showing the three memory units that specify the image format, the signals into and out of the image format controller.

Fig. 5 An example showing the timings of a fine and a coarse sampling.

Fig. 6 An example illustrating the translation of an individual scan line to achieve hexagonal sampling.

Fig. 7 Application example 1: capturing a 3x3 to 1 reduced resolution image directly without incurring image processing overhead.

Fig. 8 The sampling pattern of application example 1 with " 1 " indicating lattice sites to be sampled and blank otherwise.

Fig. 9 The resolution pattern of application example 1 with "A" indicating the resolution of the sampling point.

Fig. 10 Application example 2: using the present invention to carry out a space-variant sampling pattern.

Fig. 11 The sampling pattern of application example 2.

Fig. 12 The resolution pattern of application example 2.

Fig. 13 An illustration of the rotation coil mounted on the imager tube.

Fig. 14 An illustration of the principle of raster scan-line rotation by the application of an electromagnetic field produce by the rotation coil.

Fig. 15 Rotation of the entire raster frame by an angle a.

Fig. 16 An exemplary block diagram of a rotation and translation control unit.

This invention described hereinafter, is an image acquisition system using a camera tube imager, which can be flexibly programmed to acquire image of widely different formats in term of resolution, sampling, and orientation. The overall block diagram of the present invention is illustrated in Fig. 1. Central to the invention is a camera tube 1, also known as a tube imager, which can have any combination of beam deflection and focusing methods, i.e. electrostatic deflection/electromagnetic focus; electrostatic deflection/electrostatic focus; electromagnetic deflection/electrostatic focus or electromagnetic deflection/electromagnetic focus. The image format controller 2 of Fig. 1 coordinates the activities of all other functional units of the invention. The timing generator, 3, either locks onto external timings or generates its own timings by the oscillator 16.The timing signals generated by 3 is fed into the image format controller 2 as well as the rotation and translation control 6 and subsequently onto the deflection control unit 5 which controls the deflection of the electron beam to form the scanning raster. The timing signals refer to the line, field synchronisation (also known as the horizontal and vertical synchronisation) and blanking pulses. Beam current control and beam blanking is achieved by controlling the cathode and grid potentials which is carried out by 17 and 4, known as the beam current control unit and the grid control unit. Changing the resolution of an image sampling point is accomplished by 7 and 8 which are the resolution and focus control units respectively.Resolution control unit 7 converts digital data into analog signal and amplifies it according to the amplitude required for adjusting the focus of the electron beam. The video signal 15 is gathered at the target of the camera tube 1 and feeds into 9 which is a gating control. The gating control 9 is controlled by the image format controller 2. The gated video signal is then digitised by 10. The analog video output is available at 11 and the digital video output is available at 12. A video mixer 13 is included to produce the non-standard video signals which include the image format signals which will be described in greater depth later in the text. It takes in either the digitised video signal 12 or the analog video signals 11 and the image format signals issued by the image format controller 2 to produce a complete video signal. The rotation or translation of the image is controlled by 6 which is connected to a device 14 mounted onto the camera tube 1 as well as connected to the deflection control unit 5. The image format controller 2 also communicates with the host computer 18.

The following provides a brief description on the operation and the structure of a camera tube imager 1 which is essential to the understanding of the invention disclosed herein.

The old version tube imagers suffer several draw-backs as compared to the solidstate image sensors such as the charge-coupled device (CCD). The solid-state image sensors have claimed to eliminate the problem of image lag, image burn-in and distortions commonly associated with a tube imager. Tube imagers have however made large improvement in performance and reduction in distortion. Recently, tube imagers see the emergent of SATICON (Selenium-Arsenic-Tellurium target), HARP (High-gain Avalanche Rushing Photoconductor) or HARPICON, which have resolution and sensitivity unmatched by the solid-state image sensor. Other tube imagers like Vidicons, Newvicons, Ultricon, Plumbicons etc., have been in existence for a long time. The present invention is able to operate with tube imagers not limited to the above mentioned as long as the rasters are formed by a scanning electron beam.Functional blocks of the tube imager other than those used by the present invention will not be shown in this disclosure which is familiar to those skilled in the art of imaging.

For the purpose of describing the operation of a tube imager, the Vidicon camera is used as an example (refer to "Video Handbook" by Ru van Wezel, Newnes Technical Books, 1981). Fig. 2 shows a Vidicon tube which has electromagnetic deflection 19 and electromagnetic focus 20. This configuration is illustrative and the present invention is not limited to operate with such configuration only. The Vidicon tube converts light intensities into electrical signals by means of a photoconductive material that coats the surface of its faceplate or also known as the target 21. This semiconductor material has the characteristics of decreasing its resistance as the light intensity falling on it increases and vice versa. In those regions of the faceplate where the image is brightest, the resistance of the photoconductive surface is lowest.In effect, this surface forms a light-sensitive variable resistor.

In Fig. 2 an electron beam 22 is generated by a electron-gun 23 situated at one end of the tube. The electron beam constitutes part of the complete path for current flow. When the electron beam leaves the electron-gun and move towards the target 21 it then flows through the photoconductive surface coated on the target. The current continues, passing through the load resistor 24 and finally return to the source that powered the electron-gun.

In passing through the load resistor 24 the current develops a voltage drop that is amplified by the video amplifier 25 and becomes the video signal 15. Since the electron beam will be deflected both horizontally and vertically to produce a raster, the current flowing through the load resistor 24 will vary depending upon the instantaneous resistance of the photoconductive surface. The video signal consists of a DC (direct-current) component and an AC (alternate-current) component whose frequency and amplitude will vary depending upon the picture detail.

The tube imager provides a far greater flexibility than the solid-state image sensor in terms of the ease in changing the characteristics of the scanning rasters that form the image.

The array of photosensitive cells in a solid-state imager has fixed size and location which render modification of scanning pattern difficult. The present invention takes advantage of the flexibility of the tube imager and provides specific control over the cathode, grid, focus, deflection circuits for carrying out extremely flexible reconfiguration of the captured image.

Additional circuits are incorporated in the invention for rotation and translation of the scanning raster and hence the image captured.

Beam focusing constitutes an essential part of a tube imager. It has been traditionally required to produce a beam spot as small as possible that defines the highest resolution achievable by a particular camera tube. Any deviation from the smallest focused spot is undesirable as it degrades the resolution of the acquired image. On the contrary, this invention makes use of the defocusing of the electron beam to alter the resolution of an image which otherwise is a disadvantage in the traditional practice. There are in general two types of focusing methods, electromagnetic and electrostatic. In the case of electrostatic focusing, controlling the potential of the focusing electrodes controls the size of the electron beam impinging on the target 21.In other words, adjusting the voltage applied to the focusing electrodes adjusts the sampling area on the target and hence changes the resolution of the captured image. The resolution of each individual sampling point can then be controlled in this way. In the case of electromagnetic focusing 20, changing the current that feed into the focusing coils will also vary the spot size of the electron beam on the target as illustrated in Fig. 3. In Fig. 3 (a), current I1 30 produces the minimum spot size 31 as the electron beam 22 is being focused. Fig. 3 (b) shows current I2 32 produces a larger beam spot 33 as the electron beam is defocused. These techniques are well known in the field of electron optics (refer to "Principles of Electron Optics", by P.W. Hawkes and E. Kasper, Academic Press, 1989).

In either case, electrostatic or electromagnetic focusing, defocusing the electron beam results in the change of the sampling resolution. A focused electron beam will achieve the highest resolution and defocusing of the beam will produce lower resolution. Multiresolutional image can hence be easily acquired by changing the beam focusing together with a change of raster sampling.

The Image format controller 2 described in this invention is further illustrated in Fig. 4. The image format controller 2 includes an addresses generator 33 which receives timing signals from the timing generator 3. The address generated are to be used on a linear rectangular lattice (square lattice inclusive) 41 which consists of the x and y pointers or known as the row and column pointers 34 and 35 respectively in Fig. 7. The row and column pointers 34 and 35 are made up of digital counters. The row pointer 34 derives its count from the horizontal synchronisation signal (generated by the timing generator 3) which is also used to activate the electron beam to scan in the horizontal direction. The column pointer 35 derives its count from the vertical synchronisation signals which is also used to deflect the electron beam in the vertical direction.Both horizontal and vertical synchronisation signals are issued by the timing generator 3. The row and column pointers 34, 35 together specify the location of a finite picture element (pixel) on the image which is resided on the target 21 of the camera tube 1. Given that a tube imager is capable of producing an image of XSIZE by YSIZE which specify the maximum resolution of this particular tube imager, the address generator 33 of the present invention generates row and column pointers that can address an image of size XSIZE by YSIZE.

There are three memory units accessed by the image format controller, these are labelled as 36, 37 and 38. The resolution memory 36 and the sampling memory 37 are each of size XSIZE by YSIZE. The translation memory 38 is one-dimensional of size YSIZE which is the total number of rows in the image. Memory units 36, 37 and 38 together store the configuration of different image formats. These three memory units may be constituted by ROM (read-only-memory), PROM (Programmable ROM which further includes EPROM - Erasable PROM; EEPROM -- Electrical Erasable PROM), or memory blocks (random access memory - RAM) resided on the host system 18. Memory 37 defines the sampling pattern of an image, that is, it defines whether the electron beam is to be turned on or off at a particular target location (pixel), and whether a signal sampled at that time is to be gated into the ADC 10.Memory 36 stores the resolution of every pixel location. Memory 36 is connected to the beam-current control 17 and the grid control 4 and the resolution control unit 7. 1-bit per pixel is adequate for the sampling memory 37 to define the image sampling pattern in which "1" can be used to sample a particular location and "0" otherwise.

Before elaborating further on memory 36, the beam-current control 17 and the grid control 4 are described first. The beam current control unit 17 and the grid control 4 receive commands from the image format controller 2 and send appropriate signals to the cathode 29 and grid electrodes 28 of the camera tube 1. The grid 28 in the camera tube is usually negatively biased with respect to the cathode 29. In this way, a more negatively biased grid voltage will limit the beam current. When a voltage sufficiently negative is applied to the grid, it will turn off the beam completely and no sampling will occur.

Sending a on/off signal to the beam-current control 17 and the grid control 4 will enable them to generate the appropriate voltage to turn on or turn off the beam emitted from the cathode towards the target. The same on/off signal issued by the image format controller 2 is also used to gate out the analog video signal 15. The gating 9 is essentially an analog switch located in the path of the video signal before it is digitised. Analog video output can be tapped at 11 and the digitised video output can be tapped at 12.

Fig 5 shows a simple example to illustrate the sampling pattern stored in Memory 37. Fig. 5(a) shows the timing of a fine sampling scheme where the beam is turned on at every Ax, i.e. every increment of the x pointer results in every pixel on the target to be impinged by the electron beam. This is usually the default mode of operation for standard video format. In Fig. 5(b), a coarse sampling scheme is performed in which case the beam is only turned on at every 3Ax. In this case, the sampling memory 37 contains a value indicating the beam will be turned on only at every 3Ax. These sampling signals are also used to gate out the video signal. Effectively, a coarse resolution image is produced having a size 3x3 times less than the original image (refer to Fig. 5(a)).Fig. 5 only shows a onedimensional sampling pattern, the sampling memory which is two-dimensional essentially contains the sampling pattern of the target which is two-dimensional.

Memory 36 stores the resolution of a particular beam spot to be sampled. The resolution of a sampling point is dictated by the size of the electron beam impinging on the target. The resolution of a sample area on the target varies from dmin to dmax which is governed by the minimum and maximum size of the beam spot adjustable by the tube imager. Depending upon the number of grades of resolution required, the appropriate number of bits can be assigned to denote the resolution. Assigning 7-bit to the resolution memory 36, enables the sampling resolution to be divided into 128 (or 27 )grades, that is

Other numbers of grade can in fact be assigned in this invention which does not limit itself to the above example.This example is however, advantageous as a total of 8 bits (1 byte) per pixel is sufficient, in which 7 bits are allocated to the resolution memory 36 and 1 bit to the sampling memory 37. Such memory unit can be constructed from commercially available memory devices because of the one-byte wide data (for example, 2 x TMS44C251 make up a 8-bit wide data). The total memory space for these two memory 36 and 37 in this case will be XSIZE by YSIZE by 8-bit.

As shown in Fog. 4, the resolution control unit 7 includes a DAC (digital to analog converter) 39 which converts the digital signal that defines the size of the beam and hence the resolution of the pixel to be sampled, into an analog signal. The analog signal is amplified and corrected for linearity by the amplifier 40 before connected to the focus control 8.For a particular tube imager, if Smin (either current or voltage depending on whether electromagnetic or electrostatic focusing is used) corresponds to the minimum spot size dmin required by the focusing circuit and Smax corresponds to the maximum spot size dmax, then the signal Sd required for adjusting the focusing circuit to produce intermediate resolution d is kd, where k is a proportional constant equals to

The beam focusing characteristics of a particular imager tube is available on its specification data sheet provided by the manufacturer (for example, Philips Components XQ1270 vidicon tube). The values of 5max and Smin and hence the proportional constant k can be obtained from the data sheet.

The translation memory 38 which is only one-dimensional is addressed by the row pointers 34 only. At the beginning of every new row, the translation memory 38 will be addressed to retrieve its content. The content of the translation memory stores the amount of horizontal translation required by each particular row. The output of the translation memory 38 is fed to the rotation and translation control unit 6. This arrangement allows each individual row to be translated horizontally by a different amount. The number of bit to be used for the translation memory is dependent upon the translation resolution (in terms of fraction of Ax, see Fig. 5) and the maximum amount of translation required. Using this memory also defines the horizontal offset with respect to the reference raster frame (original frame).The translation memory 38 stores signed digital representation (for example: l's or 2's compliment) that indicates the direction of the horizontal offset.

Another use of this translation memory is in producing hexagonal sampling format where alternate rows are horizontally offset by half a sampling width of the previous scan line as shown in Fig. 6.

In using these three memory units 36, 37 and 38, a non-standard image format is first transformed into the Cartesian coordinate, and then approximated to the x-y lattice 41 and subsequently mapped onto these memory. Two examples are given here to further elaborate how different image formats can be mapped onto memory 36, 37 and 38 and reconfigure the format of image acquisition. The acquisition of coarse resolution image is used as the first example. The use of this invention to produce a space-scale Gaussian pyramid is explained. The second example describes how a complex image format similar to that of the human retina which is space-variant can be handled by this invention.

Application example 1: The technique of acquiring an image with coarse resolution is shown in Fig. 7. In Fig. 7, d1 denotes the diameter of the smallest spot size achievable by a camera tube; d2 is the diameter of a coarse sampling which encompasses 9 finest resolution samples. In this example, a 3 x 3 reduced resolution image can be easily captured without any computation. Different resolution images are captured at different frame time. Other resolution can be similarly accomplished by changing the spot size of the electron beam coupled with the appropriate sampling pattern. This method of producing coarse resolution image is more favourable over traditional methods in which reduction in image resolution is carried out by averaging m x m pixels to 1. Additional computation and image artifact are produced in the case of using traditional methods.In other words, the intensity of a coarse resolution pixel is obtained as follows:

where m is the one-dimensional resolution to be reduced, f(x,y) is the original image with (x,y) specifying the image coordinates.

There are disadvantages in generating a coarse resolution image this way. First, the process is indeed the convolution of the original image, which has the finest resolution, with a square box at regular interval specified by m. The resulting coarse resolution image appears tessellated. Secondly, although a dedicated processor such as the digital signal processing (DSP, for example, Texas Instruments TMS 320XX) chip can be used to speed up the rate of producing a reduced resolution image, it nevertheless adds cost to the overall system.

This invention is able to generate a scale-space Gaussian pyramid of any scale defined within the limit set by the maximum allowable defocused spot size of the camera tube. A scale-space Gaussian multi-resolution pyramid consists of levels that contain blurred and sub-sampled versions of the original, finest resolution image and is described by the formula:

In the above formula, t is the scale parameter greater than 0, and (x, y) E By examining the current density distribution of the electron beam, it is easy to show that the Gaussian kernel is provided by the electron beam. The distribution of the current density of the beam is usually a circular Gaussian given by J - Jo e - (r2/a2) where Jo is the current density at the centre of the beam, r is the radius, 6 is a constant denoting the spread of the beam.The impinging of the electron beam on the target can be seen as a convolution between the beam current and the charge on the target which is proportional to the light intensity formed on the target. This action therefore corresponds to the formulation of the Gaussian sampling. Controlling the spot size controls the scale parameter t and hence images of different scale can be acquired. The sampling pattern for application example 1 is shown in Fig. 8. Locations or lattice sites to be sampled are marked with "1". The rest of the locations are "0" which are not marked in the diagram for the sake of clarity. Fig. 9 shows the resolution pattern in which "A" denotes the resolution of the sampling areas.

Application example 2 is shown in Fig. 10 which illustrates a space-variant image format. This type of format is usually defined using the polar coordinate system. The concentric circles and the radial lines, all drawn in dotted lines, depict the polar coordinates as in Fig. 10. The area in the central region of this format contains the highest resolution samples, this is denoted by the dotted hexagon 42. As the distance from the centre of the image (i.e. the centre of the concentric circles and the radial lines) increases, the resolution reduces or equivalently the sampling area increases. The coarser resolution sampling is denoted by circles of bigger diameter. The centre of each of the circle denotes a point on the target to be sampled by the electron beam. The centre of the sample circle will have a corresponding Cartesian coordinate.As the target area is defined by a square lattice 41 addressed by the row 34 and column pointers 35 as shown in Fig. 4. The square lattice site nearest to the centre of the sample circle will be sampled by the electron beam. That is the corresponding location of the sampling memory 37 addressed by the row and column pointers 34 and 35 would have been written a value (eg, 1 for on and 0 for off) to indicate that the particular area is to be sampled. In other words, the content of the sampling memory 37 looks like Fig. 11, where '1' indicates lattice sites to be sampled; The rest of the memory is '0's which are shown as blank for the purpose of clarity. The location pointed by the row and column pointers in the resolution memory 36 would have been written a number equivalent to the beam spot size.Fig. 12 shows the content of the resolution memory 36 where the characters 'A' 'B' and 'C' are used to denote the three different resolution employed by this example. Resolution 'A' corresponds to the finest resolution; 'B' corresponds the intermediate resolution which defines a beam spot diameter 2.5 times that of 'A'; 'C' corresponds to the coarsest resolution which defines a beam spot diameter approximately 4.5 times that of 'A'. The rest of the memory locations are zero, which also denotes no sampling.

The scanning of the electron beam is in synchronism with the row and column pointers 34 and 35 that addresses the three image format memory units 36, 37 and 38.

The respective values retrieved from these memory units are used to turn on/off and vary the spot size of the electron beam. In this particular example, the translation memory 38 have all its location cleared to zero to denote no horizontal offset. It must be stressed that different combinations of sampling and resolution pattern are possible not limited to the above two examples elaborated.

The present invention is made to accept image formats issued by the host computer 18. The three image format memory units 36, 37 and 38 which dictate the scanning pattern of the electron beam, i.e, sampling, resolution and horizontal translation, are not limited to be constructed from ROM, EPROM or EEPROM, but can also be made up of random access memory, or memory blocks resided on the host computer 18. A program developed on the host computer transforms a user defined image format into the sampling pattern, the resolution pattern and the horizontal translation patterns. These patterns are then down loaded into memory 36, 37 and 38.

According to another aspect of the invention, rotation of the entire raster is provided by the device 14 which is an electromagnetic winding or coil mounted onto the tube imager 1 at a position after the electron beam has been deflected. Fig. 13 shows the implementation. 14 is the electromagnetic winding, 6 houses the rotation control unit which converts the signal issued by the image format controller 2 into the appropriate amount of current to energise the coil 14 so as to rotate the scanning beam to the desired angle. The principle of rotating the scanning raster beam is shown in Fig 14. 43 is the original (unrotated) scanning raster on the target 21 of the tube imager 1. 44 is the rotated raster due to the magnetic field that is created by the passing a current through the coil 14.

The amount of current energising the coil is obtained from the rotation and translation control 6. The direction of the current flowing through the coil 14 dictates the direction of rotation of the scanning raster. The rotation of the entire scanning raster is shown in Fig.

15 in which "a" denotes the angle of rotation. The centre of rotation is defined at the centre of the camera tube axis. The original rectangular raster is indicated in Fig. 15 as 45. The rotated rectangular raster is 46.

According to yet another aspect of the invention, translation of the raster is performed by adding DC signals to the original (usually has zero DC component) deflection signals. DC currents are added in the case of electromagnetic deflection circuits; while DC potentials are added in the case of electrostatic deflection circuits. Adding a DC signal to the deflection signal creates an unsymmetric deflection which is biased to one side of the deflection circuit. The amount of translation of the electron beam is proportional to the amount of DC offset signal added to the original deflection signal. Horizontal translation is achieved by adding a DC signal to the horizontal deflection signal; while vertical translation is achieved by adding a DC signal to the vertical deflection signal. The polarity of the DC offset signal dictates the direction of translation.

The rotation and translation control 6 consists of three separate paths: frame vertical translation, frame horizontal translation, and rotation. The amount of vertical translation for the entire raster frame is fed into the vertical register 47 which is then fed to the DAC as shown in Fig. 16. The output analog signal is amplified by the amplifier 48 and subsequently mixed with the original vertical deflection signal (from the timing generator 3) to produce the resultant vertical deflection signal. The frame horizontal translation signal which indicates the amount of horizontal translation desired is also first buffered in the horizontal register 49. The frame horizontal translation signal is then mixed with the line horizontal translation signal before being converted to analog signal by the DAC.The mixed frame and line horizontal translation signal is amplified 50 and subsequently added to the original horizontal deflection signal to produce the resultant horizontal deflection signal. Both the resultant vertical and horizontal deflection signals are sent to the deflection control 5 for effecting the electron beam deflection. The amount of rotation of the raster frame is stored in the rotation register 51 which is then converted to analog signal and amplified by 52. The amplified signal is sent to the rotation coil 14.

The rotation and translation image acquisition of this invention can be used on a moving platform (for examples vehicle, aircraft, ship). Sensors for sensing the orientation of the platform are readily available to the rotation and translation control of this invention.

The acquisition of images from the moving platform can then be orientated to a same reference.

The present invention advocates a communication protocol for the non-standard video signal specified by the user. This protocol encompasses the video components required for standard video communications. There are five components in the standard video signal, these are: (1) The video signal (2) Horizontal synchronizing pulses (3) Vertical synchronizing pulses (4) Horizontal blanking pulses (5) Vertical blanking pulses.

The other components required for the non-standard video signals are: (1) The sampling pattern signal (2) The resolution pattern signal (3) Horizontal translation signal for individual line (4) Horizontal translation signal for the entire frame (5) Vertical translation signal for the entire frame (6) Rotation-angle signal of the entire frame.

The sampling pattern signal is 1-bit wide denoting whether a particular image point is to be sampled. The sampling signal is first extracted and subsequently used to sample the video signal. Note that the sampling point may be larger than one pixel size depending upon its resolution and is carried by the resolution signal. The resolution signal provides information on the size of the sampling point by indicating a multiplication factor of the minimum pixel size. By indicating the resolution as a multiplication factor of the minimum sampling size, it provides generality across different video equipments. The horizontal translation component for each individual raster line is also provided. The present invention advocates this signal to be located on the front porch of the horizontal synchronizing pulses. The horizontal translation component for an individual line can be either superimposed onto the composite video signal or separately transmitted. Standard video signal, for example the RS 170 format advocates a horizontal front porch duration of 1.3 psec which is more than sufficient for the insertion of the horizontal translation signal. The signal indicating the horizontal translation of the entire picture frame is advocated to be located on the back porch of the vertical synchronizing pulses. This is so arranged such that the frame horizontal translation signal and the individual line translation signal will not be confused. The vertical translation for the entire frame is also located on the back porch of the vertical synchronising pulses apart from the horizontal frame translation signal.The frame rotational signal is also located on the back porch of the vertical sync signal as this signal only affects the entire frame but not individual raster lines. These signals may be digital or analog. These additional signals containing the video signal format is combined with the standard video components by the video mixer 13 shown in Fig. 1.

One of the immediate use of this protocol is for the display of the non-standard format video signal captured by the present invention. Standard video display equipments can be used but will not have the facility to reproduce the change in resolution, translation and rotation except for the sampling pattern. Modification to the standard video display equipments and/or the construction of a new video display equipment can be carried out by incorporating the image format controller 2, the resolution and focus control 7 and 8, the rotation and translation control 6 and the rotation coil 14. The modification and/or construction of a video display equipment for displaying images acquired by the above disclosure using the technique described comes under the embodiment of the present invention.

Claims (4)

1. A reconfigurable-format image acquisition system for acquiring images of different formats in terms of sampling pattern, sampling resolution, translation and rotation; said image acquisition system consists of a camera tube imager, image format controller and control means; said image format controller coniprises of multiple memory arrays that map the sampling pattern, sampling resolution and translation of the image to be acquired; said control means provide control for: A. adjusting the spot size of the electron beam landing on the target of the camera tube imager according to the specified sampling resolution; B. turning on and off of the electron beam according to the specified sampling pattern; C. offsetting the deflection of the electron beam according to the specified translation; D. rotating each of the horizontal electron beam scanlines according to the specified rotation.

2. A reconfigurable-format image acquisition system as claimed in claim 1 uses a coil mounted on the camera tube imager; said coil is oriented traverse to the electron beam to accomplish the rotation of the image; said coil produces an electromagnetic field to rotate the electron beam scanlines to a certain degree according to the current passing through the coil.

3. A reconfigurable-format image acquisition system as claimed in claim 2 allows the image format to be programmed by the users either directly or indirectly through the use of the host computer to alter the content of the image format memory.

4. A reconfigurable-format image acquisition system as claimed in claim 3 transform any non-standard image format into the Cartesian coordinate and then approximated to the x-y lattice and subsequently mapped onto the image format memory units.