Imaging Apparatus With Means for Cancelling Differential Offset Voltage

The present invention relates generally to image sensing apparatus, and more specifically to a difference voltage canceller for cancelling the difference between offset voltages respectively developed by analog shift registers of a charge transfer device.

Linear charge-coupled devices having several thousands of picture resolution elements are used in a satellite earth observation system that scans over a wide area of the earth surfaces to obtain high definition images. For the purpose of reducing the loss of signal intensity of charges generated by elements remote from the device output, two analog shift registers are provided, one for the odd-numbered resolution elements and the other for the even-numbered elements. In such linear CCD applications, the transfer speed of the charges, or clock frequency is determined in relation to the moving speed of the object relative to the imaging device. In addition to the attenuation problem, the outputs of the analog shift registers contain undesired offset voltages which may vary due to device variabilities as well as with operating temperature and other ambient conditions.

This situation will be explained in detail with reference to FIG. 1. V.sub.00 represents typical light intensities at different photosensitive elements of a CCD, which are alternately transferred by the odd- and even-number shift registers as an odd-numbered video output V.sub.O and an even-numbered video output V.sub.E. The odd-numbered video output V.sub.O is a sum of the odd-numbered components of video signal V.sub.00 and an offset component V.sub.1, while the even-numbered video output V.sub.E is a sum of the even-numbered components of the video signal V.sub.00 and an offset voltage component V.sub.2. By multiplexing the outputs of the analog shift registers into a time-sequential signal V.sub.C, the DC offset voltage components V.sub.1 and V.sub.2 are alternately chopped into voltage pulses. Since the voltage difference between the offset components V.sub.1 and V.sub.2 is very small, it is not critical for most applications. However, when imaging a wide area of uniform light intensity such as ocean waters, such voltage differences becomes noticeable and appears as a series of stripes. Such stripes cannot be eliminated precisely from a signal received at the earth station due to limited quantization levels of the satellite.

It is therefore an object of the present invention to eliminate the stripe pattern by reducing the difference between offset voltages developed by odd- and even-number analog shift registers of a charge transfer device.

According to the present invention, bipolar clock pulses are generated for driving the first and second shift registers of a charge transfer device. A voltage representing the difference between the offset voltages of the shift registers is generated by a band-pass filter or a differential circuit and multiplied with the bipolar clock pulses to produce a voltage of variable polarity depending on the relative value of the offset voltages. The variable polarity voltage is filtered, and a DC voltage of appropriate polarity is combined with one of the output signals of the shift registers so that the difference between the offset voltages substantially reduces to zero.

The present invention will be described in further detail with reference to the accompanying drawings, in which:

FIG. 1 is a waveform diagram showing voltages generated in a conventional charge-coupled device;

FIG. 2 is a block diagram of an image sensing apparatus according to a first embodiment of the present invention, and FIG. 2A shows a charge-coupled device with a nonlinear photosensor;

FIGS. 3A and 3B are waveform diagrams associated with the embodiment of FIG. 2, with FIG. 3A showing voltages developed when the offset voltage of the odd-number shift register of the CCD is higher than the offset voltage of the even-number shift register, and FIG. 3B showing voltages developed when the offset voltage of the odd-number shift register is lower than the offset voltage of the even-number shift register;

FIG. 4 is a block diagram of an image sensing apparatus according to a second embodiment of the present invention;

FIGS. 5A and 5B are waveform diagrams associated with the embodiment of FIG. 4, with FIG. 5A showing voltages developed when the offset voltage of the odd-number shift register is higher than the offset voltage of the even-number shift register, and FIG. 5B showing voltages developed when the offset voltage of the odd-number shift register is lower than the offset voltage of the even-number shift register; and

FIG. 6 is a block diagram of an image sensing apparatus according to a third embodiment of the present invention.

Referring now to FIG. 2, there is shown a CCD (charge-coupled device) imaging apparatus according to one embodiment of the present invention. The imaging apparatus includes a charge-coupled device 1 which consists of a photosensor array 2 which may either be of a linear type as shown in FIG. 2 or of a nonlinear type as shown at 2a, 2b in FIG. 2A, an "odd-number" analog shift register 3a and an "even-number" analog shift register 3b. The stages of "odd-number" analog shift register 3a are connected respectively to the odd-numbered photosensitive elements, or pixels of the photosensor 2 and those of "even-number" analog shift register 3b are connected respectively to the even-numbered photosensitive elements of the photosensor 2. Photogenerated charges in the odd-numbered elements of the photosensor 2 are extracted by analog shift register 3a and those generated in the even-numbered elements are extracted by analog shift register 3b. Shift registers 3a and 3b are driven by the same clock pulse of bipolar waveform supplied from a clock source 4. Photogenerated charges in the odd-numbered elements are sequentially transferred along the shift register 3a and appear as a series of odd-numbered video pulses of positive varying amplitude at the input of an amplifier 5 and those in the even-numbered elements are sequentially transferred along the shift register 3b and likewise appear as a second series of even-numbered pulses of positive varying amplitude at the input of an amplifier 6. Therefore, the video pulses from the odd-number shift register 3a are time-coincident with the positive components of the clock pulses, and those from the even-number shift register 3b are coincident with the negative components of the clock pulses.

The outputs of amplifiers 5 and 6 are coupled to summing amplifiers 7 and 8, respectively, to which an offset trimming voltage of negative polarity is applied from a diode 17 or 18 depending on the relative value of the offset components of the signals at the outputs of amplifiers 5 and 6. The outputs of summing amplifiers 7 and 8 are combined in a multiplexer 9 to produce a composite sequence of odd- and even-numbered video pulses which appear at the input of an analog-to-digital converter 10 for digital conversion.

The output of multiplexer 9 is coupled to a feedback control circuit including a band-pass filter 11 having a passband centered at the repetition frequency of the video pulses to produce a sinusoidal wave. Since the video pulses are synchronized with the clock pulse, the output of band-pass filter 11 is at the same frequency as the repetition frequency of the clock pulses.

In FIG. 3A, the output of odd-number shift register 3a is shown as containing a higher offset voltage V.sub.1 and the output of even-number shift register 3b is shown as containing a lower offset voltage V.sub.2. If these offset voltages are superimposed on a video signal, an offset-containing video output V.sub.O1 will be produced at the output of multiplexer 9 and an alternating voltage V.sub.B1 will appear at the output of band-pass filter 11. In FIG. 3B, the outputs of odd-number shift registers 3a and 3b are shown as containing a higher offset voltage V.sub.1 and a lower offset voltage V.sub.2, respectively. If these offset voltages are superimposed on a video signal, an offset-containing video signal V.sub.O2 will be produced at the output of multiplexer 9 and an alternating voltage V.sub.B2 will appear at the output of band-pass filter 11.

The amplitude of alternating voltages V.sub.B1 and V.sub.B2 represents the difference between the offset voltages V.sub.1 and V.sub.2, and their phase relationship to the clock timing indicates which one of the offset voltages is higher. If V.sub.1 is higher than V.sub.2 as shown in FIG. 3A, voltage V.sub.B1 of FIG. 3A will be advanced by 180.degree. with respect to voltage V.sub.B2 of FIG. 3B.

The output of band-pass filter 11 is applied to a phase shifter 12, where the phase of the output of band-pass filter 11 is manually adjusted so that, if V.sub.1 is higher than V.sub.2, the positive and negative components of alternating voltage V.sub.B1 respectively coincide with the positive and negative components of the clock signal V.sub.CL. Conversely, if V.sub.1 is lower than V.sub.2, the positive and negative components of alternating voltage V.sub.B2 respectively coincide with the negative and positive clock components. A multiplier 13 is provided to multiply two input waveforms, one from the phase-adjusted output of band-pass filter 11 and the other from clock source 4. The multiplication results in the generation of a series of positive sinusoidal halfwave pulses V.sub.M1 (see FIG. 3A) if V.sub.1 is higher than V.sub.2 or a series of negative sinusoidal halfwave pulses V.sub.M2 (see FIG. 3B) if V.sub.1 is lower than V.sub.2.

The output of multiplier 13 is applied to a low-pass filter 14. By the low-pass filtering action of filter 14, the sinusoidal halfwave pulses V.sub.M from multiplier 13 are smoothed into a DC voltage V.sub.L of either positive or negative polarity, which is amplified by a linear amplifier 15. The output of amplifier 15 is supplied through a unity-gain inverting amplifier 16 to the cathode of diode 17 on the one hand, and supplied direct to the cathode of diode 18 on the other. If the output of amplifier 15 is of positive polarity, the positive DC voltage will be reversed in polarity by inverting amplifier 16 and passed through diode 17 to summing amplifier 7, and if negative, it will be passed through diode 18 to summing amplifier 8.

Therefore, a negative feedback loop is established between the output of multiplexer 9 and the inputs of summing amplifiers 7 and 8 to feed back a trimming voltage of negative polarity at one of the inputs of summing amplifiers 7 and 8, depending on the relative value of the offset voltages. If V.sub.1 is higher than V.sub.2, the trimming voltage is supplied to summing amplifier 7 and acts on each odd-numbered positive video pulse from shift register 3a to reduce its amplifier so that the difference between V.sub.1 and V.sub.2 is substantially reduced to zero, producing an offset-free video output V.sub.F1 at the input of A/D converter 10. If V.sub.1 is lower than V.sub.2, the trimming voltage is supplied to summing amplifier 8 and acts on each even-numbered positive video pulse from the shift register 3b to reduce the difference between V.sub.1 and V.sub.2 to zero, producing an offset-free video output V.sub.F2 at the input of A/D converter 10.

A modified embodiment of the invention is shown in FIG. 4 in which parts corresponding to those in FIG. 2 are marked with the same numerals as used in FIG. 2. In this embodiment, the band-pass filter 11 and phase shifter 12 of FIG. 2 are replaced with 1-pixel digital delay lines or shift registers 19, 21 and digital subtractors 20 and 22 to derive a feedback signal from the output of A/D converter 10, rather than from the output of multiplexer 9. The use of the digital circuitry is advantageous for precisely determining the delay times. Digital delay line 19 delays the output of A/D converter 10 by one-pixel interval. The difference between the delayed and nondelayed outputs of A/D converter 10 is detected by digital subtractor 20 to produce a first differential signal V.sub.D1, which is passed through the second delay line 21 to the second digital subtractor 22 to which the output of the first subtractor 20 is also applied. A second differential signal V.sub.D2 appears at the input of a digital-to-analog converter 23 for coupling an analog version of the second digital differential signal to the multiplier 13 where the analog differential voltage is multiplied with the clock pulses to produce a product signal V.sub.M1. To increase the DC component of the signal V.sub.M1, the output of multiplier 13 is applied through an analog delay line 24 to a summing amplifier 25 where it is combined with the nondelayed output of multiplier 13, producing an output V.sub.M2. The output of summing amplifier 25 is filtered by the low-pass filter 14 into a DC signal V.sub.L proportional to the difference between offset voltages V.sub.1 and V.sub.2 and supplied to the linear amplifier 15 as a positive or negative trimming voltage.

As shown in FIG. 5A, if the offset voltage V.sub.1 is higher than the offset voltage V.sub.2, the analog version of the first differential signal V.sub.D1 will appear as a dotted-line waveform V.sub.D11 and the analog version of the second differential signal V.sub.D2 will appear as a solid-line waveform V.sub.D21. The multiplication with the clock pulses V.sub.CL results in a dotted-line waveform V.sub.M11 at the output of multiplier 13 and the delay-and-sum action produces a waveform V.sub.M21 of high DC content at the output of summing amplifier 25. The low-pass filtering action of filter 14 produces a positive DC voltage V.sub.L1 at the input of amplifier 15, which is inverted to negative voltage by inverter 16 and applied through diode 17 to summing amplifier 7.

If voltage V.sub.1 is lower than V.sub.2 as shown in FIG. 5B, the analog version of the first differential signal V.sub.D1 will appear as a dotted-line waveform V.sub.D12 and the analog version of the second differential signal V.sub.D2 will appear as a solid-line waveform V.sub.D22. The multiplication with the clock pulses V.sub.CL results in a dotted-line waveform V.sub.M12 at the output of multiplier 13 and the delay-and-sum action produces a waveform V.sub.M22 of high DC content at the output of summing amplifier 25. The low-pass filtering action of filter 14 produces a negative DC voltage V.sub.L2 at the input of amplifier 15, which is passed through diode 18 to summing amplifier 8.

FIG. 6 is an illustration of a further modification of the present invention which is useful for a charge-coupled device 1A in which amplifiers 5, 6 and multiplexer 9 are provided and therefore wiring access to these internal circuits is not practical. The feedback circuit of this embodiment is similar to the FIG. 2 embodiment except that it replaces the diode 18 and summing amplifier 8 of FIG. 2 with a multiplier 26 which multiplies the output of amplifier 15 with the clock pulses. Since the positive and negative pulses of the clock are respectively synchronized with the outputs of the odd- and even-number shift registers 3a and 3b, multiplication of the positive pulses of the clock with the negative voltage output of inverter 16 which is generated when V.sub.1 >V.sub.2 produces a series of negative voltage pulses synchronized with the output of odd-number shift register 3a, and multiplication of the negative pulses of the clock with the positive voltage output of inverter 16 which is generated when V.sub.1 <V.sub.2 results in a series of negative voltage pulses synchronized with the output of even-number shift register 3b.

The foregoing description shows only preferred embodiments of the present invention. Various modifications are apparent to those skilled in the art without departing from the scope of the present invention which is only limited by the appended claims.