WO1989000795A1

WO1989000795A1 - Switch arrangement for correcting picture halftone values of video signals

Info

Publication number: WO1989000795A1
Application number: PCT/DE1988/000438
Authority: WO
Inventors: Günter JOBS; Martin Molitor
Original assignee: Fraunhofer-Gesellschaft Zur Förderung Der Angewand
Priority date: 1987-07-16
Filing date: 1988-07-15
Publication date: 1989-01-26
Also published as: JPH02504338A; DE3723508C1

Abstract

The invention concerns a switch arrangement for correcting picture halftone values of the video signals of a semiconductor line image camera the diode lines of which can be illuminated by a light source and produce an image of an object located in the path of the beam in the form of a shadow on the pick-up equipment plate of the camera. The offset error induced in this type of camera by the video voltage applied is corrected in real time, together with the errors of sensitivity and the errors due to the distribution of light along a line.

Description

Circuit arrangement for gray value correction of video signals

description

Technical field

The invention relates to a circuit arrangement for gray value correction of the video signals of a semiconductor video camera according to the preamble of claim 1

Semiconductor cameras are used, among other things, in dimensional measurement technology, the most frequently used method being the so-called scanning method. In this measuring method, an object to be measured is positioned in the beam path of a parallelized or expanded white light source and the shadow image is imaged on the recording field of the semiconductor camera with an objective. As a measure of the length of an object, the position of the shadow edge between the undisturbed area of the light source and the area darkened by the measurement object is used Part used. The position of this edge is determined by the address of the image points at which a threshold value, which is defined in the intensity range and is generally at 50% of the total intensity, is intersected by the intensity curve. This value can be detected using a comparator which compares the incoming video signal with a reference voltage.

In such a measuring method, however, positional displacements of the shadow edge can occur due to non-uniformities in sensitivity, the offset error of the semiconductor image sensor and due to non-uniformities in the illumination conditions, and this can result in measurement falsifications that can be greater than the errors that occur in such measurements due to the pixel resolution of the semiconductor image sensor. So far, this error has been countered with a calibration measure in which the illumination of an undisturbed image is recorded pixel by pixel, so that after the acquisition of a second image, each pixel is related to its reference point with a correction factor by multiplication. This type of correction leads to usable results, but due to the correction calculations to be carried out there are time delays which are several times the input time for an image and which is annoying and undesirable in fast measuring systems. Since the computational effort for line scan camera systems would also be too great to relate all gray values to a reference value, such systems are considered unsatisfactory. Presentation of the invention

The invention is therefore based on the object of providing a circuit arrangement for gray value correction of the video signals of a semiconductor camera, by means of which the direct processing and without time delay Image signals of each picture element are corrected. This object is achieved according to the invention by the characterizing features of claim 1.

In the circuit arrangement according to the invention, the correction takes place during the reading out of the picture elements with a computing circuit, the gray values being raised by an offset component stored in a separate memory. This signal is then multiplied by a factor in two calculation stages in order to achieve a correction with regard to the sensitivity and the optoelectronic properties of the gray value. A memory with direct access, a read-only memory or an erasable and programmable read-only memory can be used for the necessary arithmetic circuit, in which a value can be stored for every possible input combination of gray values and reference values, which after the input values have been created and the response time at the output has expired Available. The use of such a computation for correcting the video data is the decisive feature of the invention, because classic numerical digital function generators such as adding, multiplying or dividing are not meaningfully possible for these extensive computing operations.

The fact that video signals from semiconductor image recorders are generally digitized with 8 bits is exploited in the present invention as an advantage. A further refined resolution is not advisable, since you then get into the area of optical and electronic noise. Therefore, the result of a calculation for correcting the video data need not be resolved any higher. Since semiconductor memories with sufficient storage capacity, e.g. so-called EPROMS (erasable and programmable read-only memories), which have a capacity of 64 kByte, are available on the market, calculations of second values resolved to 8 bits can be carried out in real time with any calculation rule. The term real time is understood here to mean that after a clock cycle after the input values have been applied, the result is available at the output, this process being repeated clock cycle after clock cycle. The cycle time that can be achieved by the available modules is 125 ns, which is completely sufficient to calculate the values in real time. If this cycle time is too long, faster but also more expensive blocks would have to be used. Brief description of the drawings

The invention is explained in more detail with reference to the drawing. Show it:

1 is a diagram for explaining the sensitivity and the offset error,

2 is a block diagram for a correction circuit,

Fig. 3 is a block diagram for a simplified correction circuit and

Fig. 4 shows a block diagram for a further simplified correction circuit

Matrix cameras. Best ways to carry out the invention

As the diagram in FIG. 1 shows, the light intensity of a semiconductor video camera is proportional to the applied video voltage, but is increased by the offset voltage U _o and thus falsified. It is therefore necessary to correct the gray value of a semiconductor video signal with respect to the offset error.

The circuit arrangement according to FIG. 2 is used to correct the pffset error. In this circuit arrangement, the result matrix corresponding to the selected calculation rule can be implemented with, a computer, erasable and programmable read-only memories (EPROM) and an address control. The circuit arrangement consists of the semiconductor video camera 10, which is synchronized by a clock generator 11. The semiconductor camera 10 switches its video signal to an analog-to-digital converter 12 and the synchronous signals to an address counter 13 controlled by the clock generator 11. The digitized video signal is then connected to an EPROM 14 and via an interface stage 15 to a computer 16. The address counter 13 can also be switched to synchronization by the computer 16 by a changeover switch 25 in order to be able to write to the RAMs 22, 23, 24 from the computer. The EPROM 14 is followed by two further EPROMS 17, 18. All three RAMs 22, 23, 24 can be written by computer 16 via a bus driver stage 19, 20, 21. In addition, the address counter 13 controls the EPROMS via a memory with direct access (RAM) 22, 23, 24, a switching stage 26, 27 increasing the counter reading by "1" being connected between the control to the second memory 23 and the third memory 24 is.

To correct the gray values, the computer 16 generates the required data field by program and enters this data in an address-controlled manner into the assigned RAMs. The computer 16 generates the factors necessary for the correction by first calibrating the gray values of an entire undisturbed line in the case of two takes up different degrees of illumination, whereby of the two degrees of illumination only their relationship to one another must be known, but not their absolute strengths. This has the advantage that the measurement of the irradiation strength is eliminated and that the ratio of the modulation can be determined from the ratio of the easily measured lighting currents. The respective factors are calculated by relating the values to the most driven picture element of the semiconductor camera. Subsequent darkening provides the offset values required for each pixel, the ratios of which are shown in the diagram in FIG. 1. After determining these factors, the computer 16 calculates the illumination which, in the undisturbed state, is on the semiconductor line of the video camera. For a threshold evaluation to determine the position of an edge, it is important for each location of the picture elements that the desired state of 50% shading of the originally present intensity can be determined independently of the absolute value of this illumination, provided that the digital signal is reached when it is reached a determined state is checked. The correction factors for the illumination are therefore calculated by the computer 16 in such a way that the gray value, ie the maximum possible gray value, is supplied in the case of an undisturbed state of all pixels. The results of these calculations are then entered by the computer 16 into the corresponding memory of the correction circuit.

As already mentioned, the correction of the video signals of a semiconductor camera relates to three factors. The first correction is the correction of the offset error of the picture elements, a voltage value having to be added or subtracted from the respective video signal so that the picture element generates a signal of zero volts when completely shadowed. To carry out this Correction, the values are displayed so that the first bit forms the sign bit and the other 7 bits the amount. It is therefore possible to generate a range of - 128 levels for the gray value with the 7 bits, but this range is usually only used to 10%. It is therefore possible to save memory locations in the linkage matrix by coding the offset values in only 4 bits and thus making erasable and programmable read-only memories of only 4 kB necessary. Only the sum or the difference of the values resulting from the two input values is then stored in the corresponding memory cells.

In the second correction stage, namely the EPROM 17, the sensitivity of each picture element is corrected. The sensitivity of the picture elements results from the ratio of the increase in intensity to the resulting rise in voltage of the picture element. An absolute assignment is not absolutely necessary, but only a relative one, because the picture elements must be brought to a common sensitivity. So that a free choice is possible here, too. It is necessary to increase and decrease the sensitivity of the picture elements. In the case of high-quality line image recorders, the sensitivity of the individual picture elements fluctuates by ± 10% of the standard value. Therefore, the values between 0.9 and 1.1 are required as correction factors. The factors associated with the individual picture elements are stored in the memory 23. In order to be able to carry out the increase or decrease, the values are coded in such a way that the data status of 128 becomes factor 1 and the range of factors from 0.9 to 1.1 is divided into 100 further stages. This turns 0.9 to 28 and 1.1 to 228. In the following EPROM matrix 17 are now stored under the memory cells addressed by the input words and the factors, which correspond to the respective result.

For the third correction, namely to correct the illumination, the EPROM 18 is used. In this correction, the intensity distribution of a line, which has a strongly fluctuating course, is raised to 100% illumination for further evaluation, in order to arrive at a value preselected as the end value in terms of data technology. It makes most sense to achieve the data state of 255 with 100% illumination. For this purpose, the illumination corrected for the corresponding sensitivity when reading an image is stored in the associated memory 24. The memory locations in the matrix of the EPROM 18 are assigned a result variable C such that all video values are related to the original value. The variable C results from the following formula:

C = INT ((reference value) + 0.5) t

When evaluating the video signals using the threshold value method, the value that corresponds to twice the threshold value must be used for the reference value, in this case 255. If the input value and the reference value are identical, the value 255 is present at the output. If the video value drops below the original level, the range to zero is filled in with values ranging from zero to 255, regardless of the fact that the original maximum value remains well below this limit. This makes the evaluation to determine geometric relationships, the desired equal treatment of all video values in one line is possible. This method is of particular importance by using the method proposed in patent application P 36 32 070 for expanding the resolution of a line or matrix camera.

The results of the links appear one clock cycle after the addresses have been created at the output of the EPROMS 18. This means that the addresses of the subsequent offset errors and the correction values in the memories 22, 23, 24 must be counted by one address count each than the pre-stage so that they match the incoming pixels. The resulting shift of the result by three clock cycles is irrelevant for the evaluation of the signals.

In order to obtain the correction values, the computer 16 enters the video values of an undisturbed line in its working memory. On the basis of a second line, which was emitted with an intensity that was reduced by a measurable percentage with respect to the first line, the computer 16 can calculate the sensitivity factor and the offset error of each picture element by applying a straight line equation. For measuring the intensity ratio between the first and the second image, e.g. the current through the light source can be used or a photo element which is switched as an exposure meter.

The computer 16 receives the values of the undisturbed values of a line recorded at full illumination, corrects them with the determined factors and uses the resulting values as the illumination reference for the third EPROM 18 for correction. All results are entered into memory 22, 23, 24, the addresses being generated by the separate address counter 13 and the computer 16 counting on this level after each data word written out. The clock frequency of the line camera is too high for direct writing in the camera clock and therefore the computer 16 releases the address counter 13 again so that it again runs synchronously with the line camera 10. At the same time, the memories 22, 23, 24 are switched by the computer 16 to output their data. When reading out a line, each pixel is now corrected and clock cycles shifted at the output of the circuit arrangement.

In addition to the circuit arrangement according to FIG. 2 correcting all errors, it is possible to use simplified circuits for correction. 3 shows a circuit arrangement that does not require sensitivity and offset error correction with a link matrix in torrs of an EPROM 28. In this circuit arrangement, the gray values of a line of the line camera 10 synchronized by the clock generator 11 are also fed via an analog-digital converter 12 to a link matrix, specifically to the EPROM 28. The address counter 13 controlled by the clock generator 11 and by the synchronizing pulses of the line camera controls a memory with direct access (RAM) 29 in order to switch the correction data stored therein to the EPROM 28. In addition, the digitized video data are switched to a bus driver 30 in order to be read into the RAM 29 for calibration, which drives the EPROM 28 together with the memory 29. The memory 29 can also be switched to a read position for calibrating its values. The gray values of a line are corrected by directly entering the values of a line, in which there is no object, as reference values in the memory 29 and normalizing the gray values thereon with each further readout cycle. This has the advantage that the detour via a computer is not necessary when generating the correction values, but the camera itself supplies the correction data for the memory 29.

The mode of operation is identical to that of the embodiment according to FIG. 2, namely to the third stage with the EPROM 18. At this level, the existing illumination is viewed as 100% illumination, with the result that when this limit is reached, the data status of 255 is present at the output, whereby no offset error correction is carried out and the sensitivity correction of the individual picture elements indirectly via the percentage adjustment to the Reference illumination takes place.

Entering the image data of an undisturbed line as reference values can either be triggered manually or by a computer. With the next line start after triggering the calibration process, the digitized image data are given by the bus driver 30 to the input of the memory 29 for the reference line, and at the same time the control line of this memory is switched to input. With the subsequent line start, the memory 29 operates again on the output and outputs the corresponding value of the new reference line to the calculation EPROM 28.

4 shows a further simplified correction circuit, namely a correction circuit for one Matrix camera with up to 512 x 512 pixels. For the random access memory (RAM) 29, a memory of 256 kbytes is provided, which is addressed by a column address counter 13. 1 and by a row address counter 13.2 and is dimensioned such that the content of an entire image can be stored therein. When reading out a pixel, the value of the reference pixel is applied to the calculation EPROM 28, and the correction according to a calculation table can be carried out in time with the readout. This circuit arrangement has the advantage of an accelerated illumination corrector for measurement and image data processing, and this circuit can also be implemented in an inexpensive and compact device.

Commercial usability

The circuit arrangement according to the invention is used for the correction of the video signals of a semiconductor camera.

Claims

1. Circuit arrangement for gray value correction of the video signals of a semiconductor video camera in which the diode rows can be illuminated with a light source and image an object lying in the beam path as shadow edges on the image sensor of the camera, characterized in that a computer (16) programmable with correction values with an address control Control levels (14, 17, 18) which correct the gray values of the individual picture elements with respect to the offset error, the sensitivity and the light distribution in real time within a line.

2. Circuit arrangement according to claim 1, characterized in that the semiconductor video camera (10) the video signals via an analog-digital converter (12) switched to a correction stage erasable and programmable read-only memory (EPROM 14) and switches the computer (16), which corrects the gray values in the read-only memory (EPROM 14) via a bus driver (19) and a memory (22) controlled by an address counter (13).

3. Circuit arrangement according to claim 1 or 2, characterized in that three erasable and programmable read-only memories (EPROM 14, 17, 18) connected in series from the computer (16) and from the memory (22, 23, 24) controlled by the address counter (13) ) correct the continuous video data, and that the address counter (13) controls the subsequent memories (23, 24) via a stage (26, 27) which switches the addresses by "1".

4. Circuit arrangement according to one of claims 1 to

3, characterized in that the computer (16) increases the gray values when reading the picture elements generating the video signal by the amount of the offset error and then with a factor stored in an erasable and programmable read-only memory (EPROM 17, 18) to correct the Sensitivity and / or the lighting conditions multiplied.

5. Circuit arrangement according to one of claims 1 to

4, characterized in that known memories (RAM, ROM, EPROM) are used for the correction stages, in which result values are stored for every possible input combination of gray values or reference values, which can be switched to the output within the response time after the input values have been applied .

6. Circuit arrangement according to one of claims 1 to 5, characterized in that a microprocessor used in the computer (16) interrogates the gray scale values of a line at two different degrees of illumination and by referring the signal values to the most strongly illuminated picture element the correction factors or the offset error therefrom calculated and saved in the associated memory.

7. Circuit arrangement according to one of the preceding claims, characterized in that a programmable read-only memory (PROM) or an erasable and programmable read-only memory (EPROM) is used for the arithmetic circuit, and that the gray scale values of a line in which no object appears as reference values in the memory area of the arithmetic circuit can be entered in order to normalize the gray values to it during each read cycle.