JP3627027B2 - Image detection system - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image detection system used for image blur detection.
[0002]
[Prior art]
In an image detection system, only image information of a part of a pixel area on an image sensor may be required. For example, in an electronic zoom using a video camera, conventionally, accumulated charges in an unnecessary pixel area are transferred and swept away at a high speed so that a display time for one screen is reduced to 1/60 s. It is transferred in accordance with and displayed on the entire display screen.
[0003]
[Problems to be solved by the invention]
When image blur detection is performed using an area sensor, it is necessary to repeatedly detect image information of a part of the pixel area of the area sensor within a time period of 1 s to 1 s that is a fraction of the time when the shutter is open. For this reason, the data dump time for one time must be about 1 ms. This is possible if an external high-speed A / D converter is used, but there is a problem that the cost increases.
[0004]
Also, consider applying conventional technology to this system. The conventional electronic zoom system of a video camera and this system are common in that only image information of a part of the pixel area of the area sensor is necessary. It is different in that it needs to be detected repeatedly.
[0005]
Therefore, when the prior art is applied to this system, the display time of one screen is fixed to 1/60 s determined for the TV system, and the accumulated charge in the necessary pixel area is transferred accordingly. become. However, in this system, it is not necessary to fix the display time of one screen to 1/60 s, and it is desired to finish the transfer in as short a time as possible.
[0006]
The present invention solves the above-mentioned problems in the case of applying the prior art, and at low cost without using an external high-speed A / D converter, while the shutter is opened many times. An object of the present invention is to provide an image detection system capable of performing accurate image blur detection by repeating integration and data dump.
[0007]
[Means for Solving the Problems]
In order to achieve the above object, the present invention provides an image detection system including an area sensor, an area sensor driving unit that drives the area sensor, and an area sensor output processing circuit unit that processes an output of the area sensor. Of the pixel area of the area sensor, the charge stored in the pixel area necessary for image blur detection is read at a first frequency that can be A / D converted by an A / D converter built in the microcomputer. To add the charge of multiple pixels The accumulated charges in the pixel area unnecessary for image blur detection are swept away at a second frequency higher than the first frequency.
[0008]
That is Even if the range to be read is widened, the number of pixels to be read is reduced. Also Therefore, the accumulated charge in the unnecessary pixel region is swept away at a higher speed than the reading speed of the accumulated charge in the necessary pixel region. Therefore, the time required for one integration / data dump is shorter than the case where the transfer method is not changed between the necessary area and the unnecessary area.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 1 shows a device configuration diagram of an image detection system of a camera having two area sensors 1 and 2 and a control circuit 8 for controlling them. The outputs of the area sensors 1 and 2 are used for autofocus (AF) and image blur detection as will be described later. The area sensors 1 and 2 and the monitors 3 and 4 described later are composed of photoelectric conversion elements.
[0010]
This device includes area sensors 1 and 2 having horizontal transfer registers 21a and 21b, L-shaped monitors 3 and 4 along the two sides of the area sensors 1 and 2, an AGC circuit 7, a control circuit 8, The variable gain amplifier 10, the S / H (sample hold) circuit 22, the clamp circuit 11, the output selection circuit 9, the temperature detection circuit 12, and the microcomputer μC are composed of the main elements, and each of these output buffers, An output switch is provided.
[0011]
That is, output buffers 26 and 27 for monitors 3 and 4 and output switches 5 and 6, output buffers 24 and 25 for horizontal transfer registers 21 a and 21 b, and output switches 30 and 31 are provided. In this device, the control circuit 8 is referred to as an area sensor drive unit, the gain variable amplifier 10, the S / H circuit 22, the clamp circuit 11, and the output selection circuit 9 as an area sensor output processing circuit unit. Here, the monitors 3 and 4 monitor the accumulated charges of the corresponding area sensors 1 and 2, respectively. The horizontal transfer registers 21a and 21b temporarily hold the charges of the area sensors 1 and 2 and output them serially.
[0012]
The clamp circuit 11 operates at the timing when the charge of the black reference pixel (OB) is output from the area sensors 1 and 2 and clamps the dark current voltage to a predetermined voltage. The output selection circuit 9 is common to all outputs, and the control circuit 8 selects and outputs the outputs of the area sensors 1 and 2, the outputs of the monitors 3 and 4, and the output of the temperature detection circuit 12.
[0013]
This device is formed as a one-chip IC (integrated circuit) in which each of the components excluding the microcomputer μC is provided on a single substrate. In the following, it is assumed that devices formed on this chip are formed inside, and devices not formed on this chip are formed outside. Monitor signals output from the monitors 3 and 4 are alternatively supplied to the AGC circuit 7 and the output selection circuit 9 via the output buffers 26 and 27 and the output switches 5 and 6.
[0014]
Each of the switches 5 and 6 is formed of a MOS transistor, and becomes conductive when switching signals A and B generated from the control circuit 8 are applied to the gate electrodes thereof at a low level. The monitor signal supplied to the AGC circuit 7 and the output selection circuit 9 is selected depending on which switch is turned on. That is, one of the monitor signals of the monitor 3 or 4 can be selected by the switching signal A or B. This selection of the monitor signal will be described later.
[0015]
Integration starts simultaneously in the area sensors 1 and 2 and the monitors 3 and 4. When the integration is started, the AGC circuit 7 monitors the input monitor signal for a predetermined voltage, and transmits the information to the control circuit 8 when the predetermined voltage is reached. When the control circuit 8 receives the information, it ends the integration of the area sensors 1 and 2 and notifies the external microcomputer μC that the integration has ended (hereinafter, this integration end is referred to as “natural end”). The AGC circuit 7 can be constituted by a comparator using the predetermined voltage as a reference voltage and the monitor signal as a comparison voltage, for example.
[0016]
If the monitor signal does not reach the predetermined voltage even after the lapse of the predetermined time, that is, if the information indicating that the predetermined voltage is reached from the control circuit 8 is not transmitted to the external microcomputer μC even after the lapse of the predetermined time. Instructs the control circuit 8 to forcibly terminate the integration for the area sensors 1 and 2, and the forced termination is performed.
[0017]
In both the natural termination and forced termination of the integration, when the integration is terminated, the monitor signal supplied from the output selection circuit 9 to the external microcomputer μC via the Vout terminal 46 is sent to the microcomputer μC at the timing of the termination of the integration. The A / D converter 32 built in the A / D converter 32 performs A / D conversion, and an amplification factor to be applied to the output of the area sensor is determined according to the digital value. This amplification factor is transmitted to the control circuit 8, and the amplification factor is set for the variable gain amplifier 10. Here, when the integration is naturally terminated, this amplification factor is 1. In the case of natural termination, the gain may be set to 1 without A / D converting the monitor signal to determine the gain.
[0018]
On the other hand, after the integration is completed, the outputs of the area sensors 1 and 2 are transferred to the horizontal transfer registers 21a and 21b and input to the gain variable amplifier 10 via the output buffers 24 and 25 and the switches 30 and 31, where Amplified at the set gain. The switches 30 and 31 have the same configuration as the switches 5 and 6, and the control circuit 8 generates switching signals X and Y to select the output of the area sensor 1 or 2 supplied to the variable gain amplifier 10.
[0019]
At this time, in the present embodiment, the control circuit 8 alternately selects the outputs of the two area sensors 1 and 2. The details will be described with reference to the timing chart of FIG. Here, an output from the area sensor 1 is a sensor output α, and an output from the area sensor 2 is a sensor output β. Data are output from the area sensors 1 and 2 in synchronization with the same transfer clock (not shown) generated from the control circuit 8 as shown in (a) and (b).
[0020]
The control circuit 8 switches the switches 30 and 31 that make the sensor outputs α and β conductive by switching signals X and Y that are switched at twice the frequency of the transfer clock, as shown in (c) and (d). The switches 30 and 31 are turned on when the switching signals X and Y are at a high level, and are turned off when the switching signals X and Y are at a low level. Therefore, the outputs of the area sensors 1 and 2 are alternately input to the gain variable amplifier 10 at every high level of the switching signals X and Y.
[0021]
The sensor outputs α and β amplified by the gain variable amplifier 10 are sampled and held by the S / H circuit 22. (E) shows a sampling pulse generated for this purpose. The sampled and held signal is given to the clamp circuit 11.
[0022]
Here, the voltage for the dark current is clamped to a predetermined voltage at the timing when the charge of the black reference pixel is output. The sensor outputs α and β are input to the A / D converter 32 built in the microcomputer μC via the output selection circuit 9 and the Vout terminal 46. (F) represents the waveform of the signal (output of the S / H circuit 22) supplied to the A / D converter 32 at that time. (G) simply shows the signal after A / D conversion.
[0023]
In this process, the sampling and holding is once performed using the sampling pulse in order to eliminate the influence of switching noise generated by switching the switches 30 and 31 that alternately output the switching signals X and Y at a low level to make them conductive. It is. As described above, by using the above method, the data from the two area sensors 1 and 2 output at the same timing can be taken and A / D converted while the area sensors 1 and 2 are driven with the same clock.
[0024]
In the present embodiment, the area sensors 1 and 2 in FIG. 1 are used as both sensors for image blur detection and autofocus sensors. When used as an autofocus sensor, the output of the area sensors 1 and 2 is processed by the method shown in FIGS. 2A to 2G. However, when used as an image blur detection sensor, FIG. The timing for A / D conversion is changed. Here, only differences (changes) in image blur detection will be described.
[0025]
That is, in the case of image blur detection, only one of the data of the area sensors 1 and 2 is necessary. For example, when image blur detection is performed using the output of the area sensor 1, the sensor output α is A. The A / D converter 32 is operated only at the timing given to the / D converter 32, and the A / D converter 32 is deactivated at the timing given the sensor output β. Conversely, when image blur detection is performed using the output of the area sensor 2, the A / D converter 32 is operated only at the timing when the sensor output β is given to the A / D converter 32, and the sensor output If α is given, the A / D converter 32 is deactivated.
[0026]
Instead of this method, sampling pulses are generated by switching between autofocus and image blur detection by the logic circuit shown in FIG. 3, and A / D conversion is performed with an A / D conversion start signal synchronized therewith. May be. That is, only necessary data is sampled and held, and the sampled and held data is A / D converted. As shown in FIG. 3, the logic circuit includes input terminals 33, 34, 35, and 36, NAND circuits 38 and 39, AND circuits 37 and 40, an OR circuit 41, and an output terminal 42.
[0027]
A sampling pulse α is input from the input terminal 33, and a sampling pulse β is input from the input terminal 36. Here, the sampling pulse α and the sampling pulse β will be described. The S / H circuit 22 is given sensor outputs α and β alternately by the switching signals X and Y generated by the control circuit 8 as shown in FIG.
[0028]
The sampling pulse α is a pulse generated at the timing when the sensor output α is given to the S / H circuit 22 as shown in FIG. Similarly, the sampling pulse β is a pulse generated at the timing when the sensor output β is given as shown in FIG. These sampling pulses are supplied from the control circuit 8.
[0029]
From the input terminal 35, a signal specifying whether the sensor output to be sampled and held is one or both of the two sensor outputs α and β is given. When only one sensor output α or β is used as in image blur detection, that is, when the sampling pulse α or the sampling pulse β is output from the output terminal 42 alone, a high level is input.
[0030]
Further, when both sensor outputs α and β are used as in autofocus, that is, when two sampling pulses α and β are output from the output terminal 42, a low level is input. On the other hand, in the case of image blur detection, a high level is input from the input terminal 34 when the sampling pulse α is output from the output terminal 42 alone, and a low level is input when the sampling pulse β is output from the output terminal 42 alone. A level is entered.
[0031]
For example, when a low level is input to the input terminal 35, the sampling pulse output from the output terminal 42 regardless of which signal is input to the input terminal 34 is the sampling pulse α and the sampling pulse as shown in FIG. β is synthesized. As a result, the sampling pulse is the same as that shown in FIG. 2E, and both the sensor outputs α and β are sampled and held by the S / H circuit 22.
[0032]
In addition, when a high level is input to the input terminal 35 and a high level is input to the input terminal 34, the sampling pulse α output from the output terminal 42 is a single sampling pulse α as shown in FIG. Only the sensor output α is sampled and held by the S / H circuit 22.
[0033]
Next, FIG. 5 and FIG. 6 show details of the arrangement relationship between the monitors 3 and 4 and the area sensors 1 and 2. In FIG. 5, the monitors 3 and 4 are arranged in an L shape around the area sensors 1 and 2, avoiding portions where the horizontal transfer registers 21 a and 21 b and the black reference pixels 14 and 15 exist.
[0034]
The portions where the horizontal transfer registers 21a and 21b and the black reference pixels 14 and 15 are present are avoided. If these portions exist, the monitors 3 and 4 are separated from the effective pixels of the area sensors 1 and 2, and the monitor 3 This is because the position viewed by 4 is greatly deviated from the position viewed by the actual area sensors 1 and 2. Furthermore, if the monitors 3 and 4 are arranged so as to surround the entire area sensors 1 and 2 instead of being L-shaped, the chip area is increased and costs are increased.
[0035]
In FIG. 6, the monitors 3 and 4 are arranged in a U-shape so as to avoid only the portions with the horizontal transfer registers 21 a and 21 b around the area sensors 1 and 2. The reason why the black reference pixels 14 and 15 are not avoided is that the area occupied by the black reference pixels 14 and 15 is smaller than the area occupied by the horizontal transfer registers 21a and 21b. This is because the deviation from the viewed portion is not large and is in an allowable range, and it is considered that a problem can be avoided when there is a high luminance portion in the vicinity of the black reference pixels 14 and 15. In FIG. 6, rather than cost, the purpose is to prevent overflow when a high-brightness image is formed in the vicinity of the black reference pixels 14 and 15 by reliably monitoring the image. It is said.
[0036]
FIG. 7 shows an example different from FIG. 1 of the device configuration diagram of an image detection system for a camera having two area sensors 1 and 2. The area sensors 1 and 2 in FIG. 7 are also used as both sensors for autofocus and image blur detection as in FIG. In FIG. 7, the monitors 3 and 4 of FIG. 1 are divided into 3a, 3b and 4a and 4b, and the outputs of the respective monitors 3a, 3b, 4a and 4b are output buffers 26a, 26b and 27a, 27b and switches. The signals are supplied to the AGC circuit 7 and the output selection circuit 9 through 5a, 5b and 6a, 6b. At this time, the monitor signals supplied to the AGC circuit 7 and the output selection circuit 9 are selected by the switching signals A1, A2, B1, and B2 output from the control circuit 8.
[0037]
Further, in the device shown in FIG. 7, the AGC circuit 7 is configured to have a plurality of predetermined voltages as a criterion for determining whether the integration is to be terminated naturally or forcedly, and switching of the predetermined voltage is generated from the control circuit 8. By the control signal C. For example, the control circuit 8 controls the predetermined voltage of the AGC circuit 7 in the mode in which the switching signals A1 and A2 are set to low level and the switches 5a and 5b are simultaneously turned on and the monitor signals of the monitors 3a and 3b are used simultaneously. Switching to a higher voltage by signal C.
[0038]
On the other hand, in a mode in which only one of the switches 5a and 5b is made conductive and only one of the monitors 3a and 3b is used, the predetermined voltage of the AGC circuit 7 is switched to a low voltage by the control signal C. . Even when the monitor signals of the monitors 4a and 4b are used by controlling the switches 6a and 6b, the predetermined voltage of the AGC circuit 7 is switched. Other configurations are the same as those in FIG.
[0039]
When the image detection system of FIGS. 1 and 7 is applied to image blur detection, as described above, only the data of one of the two area sensors 1 and 2 is required. Therefore, integration control is performed using the monitor signal of the monitor corresponding to the area sensor to be used. When the image detection system of FIGS. 1 and 7 is applied to autofocus, data of the two area sensors 1 and 2 are required.
[0040]
Selection of a monitor used for integration control at this time will be described with reference to a flowchart of FIG. First, in step # 5, it is determined whether or not the current integration is the first integration. If the integration is the first integration, the process proceeds to step # 10 to proceed to the first monitor (monitor 3 in FIG. 1, monitor 3a in FIG. 7). Integration control is performed using the monitor signal.
[0041]
If the current integration is not the first integration, the process proceeds to step # 15. In the previous integration result, it is determined whether there is a saturated pixel region in the effective region of the area sensors 1 and 2, and the integration is saturated. If there is a pixel area, the monitor signal of the monitor in the vicinity is used (step # 20). For example, in the image detection system of FIG. 7, it is assumed that the integration control was previously performed using the monitor signal of the monitor 3a.
[0042]
If a saturated portion is detected in the vicinity of the monitor 3b in the area sensor 1 in the previous integration result, integration control is performed using the monitor signal of the monitor 3b in this integration. Alternatively, if a saturated part is not detected in the area sensor 1 and a saturated part is detected in the vicinity of the monitor 4b in the area sensor 2, integration control is performed using the monitor signal of the monitor 4b in this integration. Do. On the other hand, if a saturated portion is not detected in the effective area of the area sensors 1 and 2 in step # 15, integration control is performed using the monitor signal of the monitor used last time as it is (step # 25).
[0043]
FIGS. 9 and 10 are schematic diagrams of optical systems when the image detection system shown in FIGS. 1 and 7 is applied to image blur detection and autofocus. FIG. 9 is a schematic diagram of an optical system for phase difference focus detection using an external light system used in a lens shutter camera or the like. The subject light in the detection sensitivity area 16 in the object scene is divided into optical paths by a pair of separator lenses 17 and projected onto a pair of area sensors 1 and 2 constituted by a CCD. Auto focus is performed by measuring the distance based on the two divided images.
[0044]
FIG. 10 is a schematic diagram of an optical system for TTL phase difference focus detection used in a single-lens reflex camera or the like. A pair of area sensors 1 composed of CCDs are obtained by dividing the light path of the light re-imaged by the condenser lens 20 positioned behind the film equivalent surface 19 on which the photographing light of the subject is imaged, by the pair of separator lenses 17. 2 is projected onto the screen. Based on the two divided images, defocus on the film surface is obtained and autofocus is performed. In either method, image blur detection is performed using an image projected on one of the two area sensors 1 or 2.
[0045]
Next, an embodiment in which the area sensors 1 and 2 of FIGS. 1 and 7 are applied to image blur detection will be described. In the case of image blur detection, only partial data of one area sensor 1 or 2 is necessary. Then, integration / data dump must be repeated many times during the short period of time when the shutter is open. Therefore, in the present embodiment, only the necessary portion of accumulated charge is read out at a data ratio that can be A / D converted by the A / D converter 32 built in the microcomputer μC, and other unnecessary portions of accumulated charge are read at high speed. The method of sweeping away is taken.
[0046]
An example of the simple control method will be described with reference to the timing chart of FIG. 11 and a more detailed configuration diagram of the area sensor of FIG. This timing chart represents one integration / data dump. In practice, the operations on this timing chart are repeatedly performed to detect image blur. Here, external signals CMP, HST, MD1, MD2, MD3, IST, RST, and CBG are given from an external microcomputer μC.
[0047]
When MD1 and CBG are at a low level, integration is possible and IST is output at a high level. In synchronization with the transition of the high level of IST, ADT is output from the control circuit 8 at a high level, and unnecessary charges accumulated in the pixels are discharged. During this period, three timing pulses P1, P2, and P3 are sequentially generated as MD3 from the microcomputer μC to the control circuit, and various modes of the image detection system of FIGS. 1 and 7 are set using MD2 as a data signal. Here, the system of FIG. 1 will be described as a representative.
[0048]
First, the contents to be set are selection of area sensors 1 and 2, selection of H / L gain (sensitivity), and selection of addition / non-addition of pixel data described later. After these selections are made, IST is output at a low level, so that the control circuit 8 starts integration of the two area sensors 1 and 2 and the respective monitors 3 and 4. When the integration is started, the monitor signal of the monitor 3 or 4 corresponding to the area sensor 1 or 2 previously selected as the Vout signal is output from the Vout terminal 46 via the output selection circuit 9.
[0049]
At the same time, the monitor signal is supplied to the AGC circuit 7 from the switch 5 or 6. When the monitor signal reaches a predetermined voltage within a predetermined time, the AGC circuit 7 transmits this information to the control circuit 8, and the control circuit 8 ends the integration of the area sensors 1 and 2 (natural end). At the same time, the control circuit 8 outputs ADT at a low level and notifies the external microcomputer μC that the integration has been completed. After completing the integration, the control circuit 8 internally generates a shift pulse, and transfers the charges accumulated in the pixels 29 of the area sensors 1 and 2 to the vertical transfer register 18 through the shift gate 13.
[0050]
However, if the monitor signal of the monitor 3 or 4 does not reach the predetermined voltage of the AGC circuit 7 even after the lapse of a predetermined time, that is, the ADT is not output at a low level from the control circuit 8 to the external microcomputer μC even after the lapse of the predetermined time. The external microcomputer μC forcibly terminates integration by outputting CBG at a high level. Accordingly, ADT is output at a low level.
[0051]
At the same time as the integration is completed, the monitor signal of the monitor 3 or 4 output from the Vout terminal 46 is A / D converted by the A / D converter 32. Then, the gain of the variable gain amplifier 10 with respect to the sensor outputs α and β is determined according to the result of the A / D conversion. The monitor signal is continuously output from the Vout terminal 46 until the output selection circuit 9 selects the output from the temperature detection circuit 12.
[0052]
After the integration is completed, if there is an unnecessary horizontal line in the selected area sensor 1 or 2 first, the microcomputer μC outputs HST at a high level. Then, a vertical transfer clock Vcp is generated from the control circuit 8 and the charge is transferred vertically.
[0053]
At this time, Cout synchronized with the internal pixel transfer clock (vertical transfer clock Vcp, horizontal transfer clock Hcp) is output at high speed from the control circuit 8 as the HST transitions to a high level. That is, the vertical transfer clock Vcp also becomes high speed, and unnecessary charges on unnecessary horizontal lines are vertically transferred and discharged at high speed. As in the Cout output in this case, the fact that many pulses are continuously output is represented by a square by connecting both the high level and the low level with straight lines on the timing chart.
[0054]
While an unnecessary horizontal line is vertically transferred, the control circuit 8 outputs RCG at a high level and opens the register clear gate 23. Therefore, unnecessary charges of unnecessary pixels are discarded by being discharged to OD28. The counter 43 of the microcomputer μC counts unnecessary horizontal lines. When the counting is completed, CMP is output from the microcomputer μC to a high level, the control circuit 8 stops the vertical transfer, and closes the register clear gate 23.
[0055]
Next, in order to set the previously determined gain in the variable gain amplifier 10, the microcomputer μC first generates a timing pulse P4 based on the MD3 signal. Accordingly, the microcomputer μC inputs IST and MD2 as data signals to the control circuit 8, and the control circuit 8 sets the gain information set by the IST and MD2 in the gain variable amplifier 10. In this case, the contents of IST and MD2 are amplification factors to be set.
[0056]
Thereafter, when the microcomputer μC outputs MD1 at a high level, the area sensors 1 and 2 are set in the reading mode. Then, when the microcomputer μC generates an RST pulse, the area sensors 1 and 2 start reading.
[0057]
Along with the generation of the RST pulse, the control circuit 8 generates one vertical transfer clock Vcp, vertically transfers the charges of each horizontal line by one line, and the microcomputer μC resets the built-in counter 43. Further, the control circuit 8 supplies the horizontal transfer clock Hcp to the horizontal transfer register 21 to horizontally transfer the charges in the horizontal transfer register 21. At this time, as described with reference to FIG. 2, the outputs from the two area sensors 1 and 2 are alternately input to the variable gain amplifier 10. That is, the charges of the corresponding pixels of the area sensors 1 and 2 are sequentially input alternately.
[0058]
The sensor outputs α and β are input to the clamp circuit 11 via the S / H circuit 22. In the clamp circuit 11, the CBG is low level from the microcomputer μC at the timing when the charge of the black reference pixel is output. In response to this, the clamp circuit 11 operates to clamp the dark current voltage to a predetermined voltage.
[0059]
After the charge of the black reference pixel is clamped, the sensor outputs α and β are given to the output selection circuit 9. When selected by the output selection circuit 9, it is output from the Vout terminal 46, but the output at that time is every half of the period in which the data of one pixel is output from each area sensor 1, 2. The sensor output α and the sensor output β are alternately present (see FIG. 2F).
[0060]
In the case of image blur detection, as described above, only one sensor output α or β is necessary, so that the A / D converter 32 inputs the sensor output α or β to the A / D converter 32. A / D conversion is performed at the timing. The A / D converted sensor output data is used for image blur detection.
[0061]
The image blur detection method will be described later with reference to FIG. In the above embodiment, the outputs of the two area sensors 1 and 2 are alternately taken out. However, only the output of one area sensor 1 or 2 may be selected at the stage of sample and hold. Only one sensor output α or β is input to the A / D converter 32. When only one output is sampled and held, a high level may be input to the input terminal 35 in the logic circuit of FIG.
[0062]
Returning to FIG. 11, after the end of the above-described clamping, the microcomputer μC sets CBG to high level. If there is an unnecessary pixel immediately after the black reference pixel, the microcomputer μC sets the number of unnecessary pixels in the counter 43 shown in FIG. 1, and the microcomputer μC removes the HST in order to sweep away the charges of the unnecessary pixels at high speed. Set to high level.
[0063]
Accordingly, the control circuit 8 similarly generates a horizontal transfer clock Hcp at high speed, and charges of unnecessary pixels are horizontally transferred at high speed and discharged. The Cout clock is output at high speed from the Cout terminal. The charges of the unnecessary pixels are eventually discarded because they are not selected by the output selection circuit 9 (and therefore are not led to the Vout terminal 46).
[0064]
The counter 43 of the microcomputer μC counts Cout clocks for unnecessary pixels, and after the count ends, the microcomputer μC sets CMP to a low level. After the transition of the CMP low level, the microcomputer μC changes HST to low level and further sets IST to high level. The control circuit 8 generates the horizontal transfer clock Hcp at a low speed with the transition of the low level of the HST, and outputs an effective pixel signal (valid signal) from the Vout terminal 46 with the transition of the high level of the IST.
[0065]
When the counter 43 of the microcomputer μC finishes counting the number of effective pixels, the microcomputer μC outputs CMP at a high level. Here, if unnecessary pixels still remain on the horizontal line, the microcomputer μC sets the number of unnecessary pixels in the counter 43 and sets HST to a high level, so that the charges of the unnecessary pixels are internally horizontal at high speed. It is transferred, discharged and discarded. Further, immediately after the transition of the HST to the high level, the microcomputer μC sets the IST to the low level, while the control circuit 8 outputs the reference voltage (REF) from the Vout terminal 46 instead of the unnecessary pixel output.
[0066]
After the count of the remaining unnecessary pixels is completed, the microcomputer μC outputs CMP at a low level, and the control circuit 8 ends the horizontal transfer. Then, to read the next horizontal line, the microcomputer μC generates an RST pulse. Accordingly, the control circuit 8 generates one vertical transfer clock Vcp, vertically transfers the charges of each horizontal line by one line, and repeats the above-described horizontal line reading operation. The CCDs of the area sensors 1 and 2 according to the present embodiment use the interline transfer (IT) method, but of course, the frame interline transfer (FIT) method or the frame transfer (FT) method may be used.
[0067]
Further, in the present embodiment, when setting various modes of the image detection system, the addition / non-addition of pixel data is selected. Before describing the selection of addition / non-addition of pixel data, the significance of the addition will be described. First, in the case of image blur detection as described above, since integration and data dump must be repeated many times in a short time, the wider the pixel area to be read, the longer it takes, and the response in image blur detection Becomes worse. Therefore, in order to enlarge the readable pixel area as much as possible and reduce the number of pixels to be read as much as possible, a method of adding the charge of the adjacent pixel to the charge of the read pixel is introduced.
[0068]
Specifically, when the detected area is small for a low-frequency subject, a sufficient contrast cannot be obtained, so that a correct comparison cannot be performed and a correct image blur amount cannot be detected. In order to solve this problem, it is necessary to read out a wider area. However, if the wide area is simply read out, the time required for one reading is increased, and sufficient image blur detection and image blur correction cannot be performed.
[0069]
Further, it is sufficient to read out pixels at a high speed by the amount corresponding to a wider read area, that is, by increasing the number of read pixels. I can't do anything. Therefore, in this embodiment, by adding the charges of a plurality of pixels during charge transfer, the range to be read is widened, but the number of pixels to be read is not changed. In addition, since charges of a plurality of pixels are added, there is an advantage that sufficient charges can be obtained in a short integration time even under low luminance.
[0070]
Details of the driving method at this time will be described with reference to the timing charts of FIGS. 13 and 14 and the configuration diagram of the area sensor of FIG. FIG. 13A is a timing chart of the vertical non-addition mode. Here, one pulse of the vertical transfer clock Vcp is input, the charge of each horizontal line is vertically transferred by one line, and each time the charge of one horizontal line is transferred to the horizontal transfer register 21, the pixel of one horizontal line is transferred. The horizontal transfer clock Hcp for several minutes is continuously input to transfer charges horizontally.
[0071]
On the other hand, in the vertical direction N pixel addition mode (N = 2 in this embodiment) of FIG. 13B, N pulses of the vertical transfer clock Vcp are input, and charges for N horizontal lines are input to the horizontal transfer register 21. By inputting, the charges for N pixels adjacent in the vertical transfer direction in the horizontal transfer register 21 are added, and then the horizontal transfer clock Hcp is input.
[0072]
FIG. 14A is a timing chart of the horizontal non-addition mode. Here, every time one pulse of the horizontal transfer clock Hcp is inputted, a sampling pulse is inputted and sampled and held for every charge of one pixel. On the other hand, in the horizontal N pixel addition mode in FIG. 14B (N = 2 in this embodiment), the horizontal transfer clock Hcp is input at a frequency N times that in FIG. The frequency is not changed.
[0073]
Therefore, since the sample / hold is performed every time the charge for N pixels is inputted in the S / H circuit 22, the charge for N pixels in the horizontal direction is added. Therefore, the transfer clock of the data input to the A / D converter 32 does not change, but the charges for N pixels in the horizontal direction are added and input. Selection of addition / non-addition will be described with reference to the flowchart of FIG. First, in step # 35, it is determined whether or not the current integration is the first integration. As a result, if it is the first integration, it is not known whether the data is reliable. Read out with.
[0074]
If the current integration is not the first integration, the process proceeds to step # 40. If the previous detection result is reliable, the process proceeds to step # 65, and the previous read mode is continued to perform the integration. If the previous detection result is unreliable, the process proceeds to step # 45 to determine whether the previous integration was performed under low luminance. If the integration was performed under low luminance, step # 45 is performed. Proceeding to 55, the current reading is performed in the addition mode.
[0075]
If the previous integration was not performed under low luminance, the process proceeds to step # 50, and if the previous readout mode is the addition mode, the current readout is performed in the non-addition mode (step # 60), and the previous readout is performed. If the mode is the non-addition mode, the current reading is performed in the addition mode (step # 55). By controlling in this way, it is possible to read out the data by the current integration in the optimum reading mode based on the previous detection result and brightness. Further, in the addition mode, the mode of addition only in the vertical direction, addition only in the horizontal direction, and addition in both directions is switched according to the contrast, frequency, control sequence, and the like of the subject.
[0076]
Next, a case where the area sensors 1 and 2 of the present embodiment are applied to autofocus will be described. When the area sensors 1 and 2 are applied to autofocus, data of a specific partial area of the two area sensors 1 and 2 is read out from data obtained by one integration. That is, it is used in the line sensor mode.
[0077]
The size of one pixel of a normal area sensor is only a fraction of the pixel size of a line sensor used for autofocus, and the sensitivity is insufficient for a low-luminance subject. For this reason, in the present embodiment, the charges accumulated for several pixels of the area sensors 1 and 2 are added to the charges for N pixels in the vertical direction according to the timing chart of FIG. Ensure sensitivity.
[0078]
The driving method at this time will be described with reference to the block diagram of FIG. 12 and the timing chart of FIG. First, a timing chart in the normal area sensor mode is shown in FIG. As an integration start operation, a shift pulse is input, the shift gate 13 is opened, and charges accumulated in each pixel 29 before integration are discharged to the vertical transfer register 18. This starts integration.
[0079]
Next, the register clear gate 23 assigned to the horizontal transfer register 21 is opened by setting RCG to a high level. In this way, the charges transferred to the horizontal transfer register 21 can be discharged to the OD 28. Then, in order to vertically transfer charges for all horizontal lines, a vertical transfer clock Vcp is input, and charges initially accumulated in each pixel 29 are discharged to OD 28 (integration clear time). Thereafter, RCG is set to low level, and the register clear gate 23 is closed.
[0080]
When it is detected using the monitors 3 and 4 that sufficient integration has been performed, the integration is naturally terminated (of course, the integration may be forcibly terminated). Thereafter, a shift pulse is input and the shift gate 13 is opened to transfer the charge of each pixel 29 to the vertical transfer register 18. Then, after one pulse of the vertical transfer clock Vcp is input to transfer the charges of the horizontal pixel column for one line to the horizontal transfer register 21, the horizontal transfer clock Hcp is continuously input to drive and read the horizontal transfer register 21. Thereafter, by repeating this operation, the charges of all the pixels 29 are read out.
[0081]
FIG. 16B shows a timing chart when the area sensors 1 and 2 are switched to the line sensor mode for auto-focusing with respect to the normal area sensor mode. Until the integration end operation, it is the same as the normal area sensor mode. Thereafter, when the vertical transfer is performed by inputting the vertical transfer clock Vcp, unnecessary charges in the horizontal pixel column not used as a line sensor are simultaneously set to RCG at a high level, the register clear gate 23 is opened, and the horizontal transfer register 21 is opened. To OD28.
[0082]
For a horizontal pixel column used as a line sensor, the vertical addition mode shown in FIG. 13B is used to set RCG to low level, close the register clear gate 23, and configure the line sensor. By inputting the vertical transfer clocks Vcp for the required number of pixels, the charges of the pixels are combined (added) in the horizontal transfer register 21. Then, the horizontal transfer clock Hcp is input to drive the horizontal transfer register 21 and read it as a line sensor. By driving in this way, the area sensors 1 and 2 can be used as line sensors.
[0083]
Further, in autofocus, it is necessary to use the two area sensors 1 and 2 as the line sensor mode and to read out data obtained from the two area sensors 1 and 2 by one integration. As for the driving method at that time, as described above with reference to the timing chart of FIG. 2, the two sensor outputs α and β are alternately supplied to the A / D converter 32 to perform A / D conversion.
[0084]
When the area sensors 1 and 2 in FIGS. 1 and 7 are used as an autofocus sensor in the line sensor mode, the part used as a line sensor is monitored as much as possible in this embodiment so that the part used as a line sensor does not overflow. It is set to 3 and 4 neighborhoods. Furthermore, the area sensors 1 and 2 are also used for image blur detection.
[0085]
Here, the relationship between the part used for autofocus and the part used for image blur detection (selected from the entire area sensor) is important with respect to the optical axis center of the optical system. The parts used for the sensor and the optical axis center of the optical system will be described with reference to FIGS. Here, the system of FIG. 7 will be described as a representative.
[0086]
In the first example of FIG. 17A, the autofocus area 44 is set in a portion close to the monitors 3b and 4b, and the optical axis center 45 of the optical system is not the center of the autofocus area 44 but the entire area sensors 1 and 2. It is set at the center. Such an arrangement is inconvenient because the center of the autofocus area 44 is not at the center of the screen, but if the detection area of the entire area sensors 1 and 2 is not so large relative to the entire screen, the autofocus area 44 The deviation from the screen is not so large and is acceptable. In addition, when image blur is detected, the probability of capturing the main subject is increased.
[0087]
In the second example of FIG. 17B, the autofocus area 44 is set near the monitors 3b and 4b, and the optical axis center 45 of the optical system is set at the center of the autofocus area 44. Such an arrangement has an autofocus area 44 at the center of the screen, which is the same as a normal autofocus sensor. On the other hand, the image blur detection area deviates from the center of the screen and the probability of catching the main subject is low, but it is effective because the possibility of catching a stationary object in the background is high. At this time, it is necessary to set so that the image blur detection area is not empty or only on the ground.
[0088]
In the third example of FIG. 17C, the pixel arrangement is the same as in FIG. 17B, but the positions of the horizontal transfer registers 21a, 21b are changed depending on the chip layout of the ICs of the area sensors 1, 2. is there. When the horizontal transfer registers 21a and 21b are arranged in this way, the same reading method as in FIG. In this case, vertical transfer is performed for each line, pixel charges are added during horizontal readout, and necessary data is selectively stored during A / D conversion.
[0089]
In the fourth example of FIG. 17D, the optical axis center 45 of the optical system is the center of the area sensors 1 and 2, and the centers of the autofocus area 44 and the image blur detection area are arranged at this optical axis center 45. . With this arrangement, the autofocus area 44 or the image blur detection area shown in FIGS. 17A and 17B does not deviate from the optical axis center 45, but the autofocus area 44 and the monitor are separated. For this reason, it becomes difficult to perform integral control effectively using the monitor signal.
[0090]
Therefore, in this embodiment, the first integration is performed by using a monitor voltage (monitor signal) of a certain monitor, but in the second and subsequent integrations, the previous integration time and the current monitor voltage of the monitor used last time are used. Control integration based on. A specific example of the control method at this time will be described with reference to FIG. First, in the first integration, the area sensors 1 and 2 are integrated until the time when the monitor voltage reaches the predetermined voltage V1 of the AGC circuit 7 by the above-described method, and the process is naturally terminated (the integration time at this time is t1).
[0091]
For example, at this time, it is assumed that accurate integration control is not performed by the monitor output, and it is determined that the integration amount in the area sensors 1 and 2 is excessive. Since the rate of change of the integration amount of the area sensor is considered to be constant, the optimum integration time t3 can be obtained from the time t1 and the excess integration amount.
[0092]
As with the area sensor, the rate of change of the monitor voltage is considered to be constant, so the relationship between time and the monitor voltage is a straight line shown in FIG. From this straight line, the monitor voltage at the time t3 in the first integration, that is, the optimum monitor voltage V3 at the end of integration is known. In the second integration, the integration is performed until the time t required for the monitor voltage to reach V3. It can be seen from FIG. 18 that the time t can be obtained using the following equation (equation (1)).
[0093]
t = t3 × (2 × t2 / t1) (1)
[0094]
Here, t2 is a time until the monitor voltage reaches a voltage V2 that is ½ of the predetermined voltage V1 in the second integration. The time t2 is monitored using the AGC circuit 7. That is, a predetermined voltage V1 and a voltage V2 that is 1/2 of the predetermined voltage are set in the AGC circuit 7, and the time t2 when the monitor voltage reaches V2 is monitored. The microcomputer μC calculates t from the equation (1), and forcibly terminates integration at time t. In the third and subsequent integrations, integration control is performed in the same manner as in the second time with the previous integration time set to t1.
[0095]
FIG. 19 shows an integration control method in which the method of FIG. The process until the optimum voltage V3 is obtained by setting the predetermined voltage V1 in the AGC circuit 7 and performing integration is the same as in the case of FIG. Here, with this V3 as a predetermined voltage at the end of integration, the microcomputer μC monitors the monitor voltage in the next integration, and the integration is controlled by forcibly terminating when the monitor voltage reaches V3.
[0096]
Further, the monitor natural end voltage V3 is inputted to the AGC circuit 7 from an external D / A converter, and the above-mentioned optimum monitor voltage V3 is set in the AGC circuit 7 as a predetermined voltage for the natural end of integration from the outside. Then, the second and subsequent integrations can be controlled using the natural termination function.
[0097]
Next, selection of the autofocus area will be described with reference to a simple flowchart of FIG. First, in order to prevent the area used for autofocus from being saturated, it is checked in step # 70 whether the area near the monitors 3 and 4 can be used. In this method, the contrast of an area to be used for autofocus is obtained, and it is determined whether there is sufficient contrast for performing autofocus. If it is determined that there is sufficient contrast, the process proceeds to step # 75 to use this area.
[0098]
If it is determined that there is no contrast, the process proceeds to step # 80, the detection area is moved to the next area, and the process proceeds to step # 85 to check the contrast. If it is determined that there is contrast in this area, the process proceeds to step # 90 to use this area. If it is determined that the area is not yet sufficiently contrasted, the process returns to step # 80, the area is further moved, the process proceeds to step # 85, the contrast calculation is repeated, and an area having sufficient contrast is searched for. Go to and determine the autofocus area.
[0099]
In the second and subsequent autofocus operations, the detection area is continuously used in the previous detection area. If a low contrast is detected in this area (which means an unreliable state), the detection area is moved to this vicinity and is continuously executed. Yes. Next, one method for performing image blur detection by selecting an image blur detection area will be described with reference to a simple flowchart of FIG. 21 and a diagram showing an image blur detection sequence of FIG. The lattice-shaped part in FIG. 22 shows the light receiving part of the sensor.
[0100]
First, after the autofocus is completed, the process proceeds to step # 100 to determine whether panning has occurred. If there is no panning, it is considered that the main subject is present in the autofocused area, so the process proceeds to step # 105 and within the autofocus area. A region with sufficient contrast (for several pixels) is searched, and if a region with contrast is found, the process proceeds to step # 115 and the region is shown in FIG. 2 As shown in (a), it is used as a reference part for image blur amount calculation. FIG. 22A shows the reference unit 100 in this case. At this time, the reference unit 100 specifies where in the area sensor 1 or 2 is located.
[0101]
In order to perform image blur detection, integration and data dump are further repeated while the shutter is open. When the data obtained by the next integration is read, the area to be read is read as a reference part 110, which is a wide area centering on the previously determined reference part 100 as shown in FIG. The reference part is provided here by limiting the region where the reference part can exist, thereby reducing the probability of recognizing the moved reference part as a region having a different contrast, and reading only the reference part. This is because it can be read out faster than the whole.
[0102]
Then, the region 120 having the highest correlation with the reference unit 100 is detected in the reference unit 110, the position in the area sensor 1 or 2 is specified, the amount of movement from the reference unit is calculated, and this is the image blur. As an amount, image blur correction is performed based on this. Further integration is performed, and this time, as shown in FIG. 22C, this area is read as a reference area 111 that is one area wider than the area 120 having the highest correlation with the reference portion. Then, the region 121 having the highest correlation with the region 120 in the reference unit 111 is detected, the position in the area sensor 1 or 2 is specified, the movement amount from the region 120 is calculated, and this is calculated as the image blur amount. Based on this, image blur correction is performed.
[0103]
Further, integration is performed, and as shown in FIG. 22 (d), a region that is slightly wider around the region 121 is read as the reference unit 112, and the region 122 having the highest correlation with the region 121 is detected in the reference unit. Alternatively, the position in 2 is specified, the amount of movement from the region 121 is calculated, and this is used as the amount of image blur, and image blur correction is further performed based on this. The same operation is repeated until the shutter closes to correct image blur.
[0104]
Returning to FIG. 21, if it is determined in step # 100 that panning has occurred, it is considered that there is no main subject in the autofocused area, so the process proceeds to step # 110, where the contrast is increased from other than the autofocus area. A certain area is searched, and if a contrast area is found, the process proceeds to step # 115, and the image blur detection is performed as described above using the area as a reference unit for detecting the image blur.
[0105]
Here, when panning is performed, it is considered that the main subject is deviated from this region. Therefore, a region having contrast is searched for except for the region where autofocus is performed. Of course, an area having contrast may be sequentially searched without removing this area.
[0106]
In the panning determination method, the sensor output can be obtained from the correlation of the light distribution pattern at predetermined time intervals, as in the case of camera shake detection. Obtain the temporal deviation amount of the light distribution pattern at a predetermined time interval of the sensor output used for camera shake detection. The panning amount can be detected based on the deviation amount.
[0107]
As another method, an automatic focus detection operation can be used. In the automatic focus detection operation, if the defocus amount detected by the focus detection operation is equal to or less than a predetermined value, it is determined that the subject is a stationary subject, and if it is equal to or greater than a predetermined value, it is determined that panning has been performed. Then, the AF lock mode in which the automatic focus detection operation is stopped is entered. If the detected defocus amount is between the two predetermined values, it is determined as a moving subject and the continuous AF mode is entered. It is also possible to determine panning based on this operation.
[0108]
【The invention's effect】
According to the present invention, in an image detection system having an area sensor, the time required for one integration / data dump is shortened without requiring a high-speed video A / D converter. Therefore, it is possible to repeat integration and data dump a number of times while the shutter is open without incurring extra cost, and to perform accurate image blur detection.
[Brief description of the drawings]
FIG. 1 is a device configuration diagram of an image detection system having two area sensors and a control circuit for controlling them.
FIG. 2 is a timing chart related to the output from the area sensor of FIG.
FIG. 3 is a diagram showing a logic circuit for adjusting a sampling pulse.
FIG. 4 3 The figure which showed the specific example of the sampling pulse generated from the logic circuit.
FIG. 5 is a diagram showing details of the arrangement relationship between a monitor and an area sensor.
FIG. 6 is a diagram showing another example of details of the positional relationship between the monitor and the area sensor.
FIG. 7 is an example different from FIG. 1 of the device configuration diagram of an image detection system having two area sensors and a control circuit for controlling the two area sensors.
FIG. 8 is a flowchart for selecting a monitor in the case of autofocus.
FIG. 9 is a schematic diagram of an optical system to which the image detection system of FIGS. 1 and 7 is applied.
FIG. 10 is a schematic diagram of another optical system to which the image detection system of FIGS. 1 and 7 is applied.
FIG. 11 is a timing chart of an image blur detection control method.
FIG. 12 is a detailed configuration diagram of an area sensor.
FIGS. 13A and 13B are timing charts in the case of (a) non-addition in the vertical direction and (b) addition of charges for two pixels in the vertical direction.
14A and 14B are timing charts in the case of (a) non-addition in the horizontal direction and (b) addition of charges for two pixels in the horizontal direction.
FIG. 15 is a flowchart of selection between addition and non-addition.
FIG. 16 is a timing chart of (a) normal area sensor mode and (b) line sensor mode in the case of autofocus.
FIGS. 17A and 17B are diagrams showing an autofocus area and an optical axis center in an area sensor of a first example, a second example, a second example, a third example, and a fourth example; FIG.
FIG. 18 is a diagram showing the relationship between the integration time and the monitor voltage when the integration of the area sensor is controlled by time.
FIG. 19 is a diagram showing the relationship between the integration time and the monitor voltage when the integration of the area sensor is controlled by the monitor voltage.
FIG. 20 is a flowchart of selection of an autofocus area.
FIG. 21 is a flowchart of selection of an image blur detection area.
FIG. 22 is a diagram illustrating an example of a sequence obtained by (a) first (b) second (c) third (d) fourth integration during image blur detection.
[Explanation of symbols]
1, 2 Area sensor
3, 4 monitor
3a, 3b monitor
4a, 4b monitor
5, 6 Monitor output switch
7 AGC circuit
8 Control circuit
9 Output selection circuit
10 Gain variable amplifier
11 Clamp circuit
μC microcomputer
14, 15 Black reference pixel
18 Vertical transfer register
21 Horizontal transfer register
21a, 21b Horizontal transfer register
22 S / H circuit
23 Register clear gate
30, 31 Area sensor output switch
32 A / D converter
46 Vout terminal

Claims (1)

In an image detection system having an area sensor, an area sensor driving unit that drives the area sensor, and an area sensor output processing circuit unit that processes the output of the area sensor,
Of the pixel area of the area sensor,
The accumulated charge in the pixel area necessary for image blur detection is read out at a first frequency that can be A / D converted by an A / D converter built in the microcomputer, and the charges of a plurality of pixels are added .
An image detection system, wherein stored charges in a pixel area unnecessary for image blur detection are swept away at a second frequency higher than the first frequency.

Images