MXPA99007268A - Analog circuit for an autofocus microscope system - Google Patents

Abstract

An analog circuit for an autofocus microscope system measures a degree of focus of an object directly from the video signal of a microscope CCD camera. The circuit then returns an index to a host computer for the purpose of adjusting the position of the microscope's objective lens to bring the object in focus. Best focus ins found by comparing indices at several different vertical positions. The criterion adopted for determining the degree of focus is derived from the energy distribution of the video signal spectrum. The signal passes through a highpass filter, a gate (to remove scanning artifacts), a squarer and an integrator. The high frequency energy of the video spectrum is a maximum at best focus and as the optics defocus, the distribution shifts to lower frequencies. Low cost, real time autofocus is achieved with the analog circuitry of this invention, replacing more expensive dedicated real time image processing hardware.

Description

ANALOGUE CIRCUIT FOR AN AUTOMATIC APPROACH MICROSCOPE SYSTEM TECHNICAL FIELD The invention relates to the field of microscopy, in which the focus of an image, observed with a microscope, is automatically adjusted. Such automatic adjustment of the focus of the microscope is called "autofocus". The invention is implemented in an automatic focusing system. Specifically, the invention relates to the incorporation of an analog circuit that performs a transfer function attributable to the microscope lenses of the autofocus system, eliminates scanning artifacts that impair the autofocus function, and corrects the instability of lighting.

PREVIOUS INVENTIONS Autofocus is essential in automated microscopy to overcome the problems of mechanical instability, the irregularity of slides and coverslips, the movement of live specimens and the effects of thermal expansion. Autofocus can overcome these limitations and allows accurate measurements, which are irreproducible in automated quantitative microscopy. There will be many experiments that benefit from the automatic approach, among whose examples it is indispensable to include the following: (a) the exploration of large areas with high resolution, where the depth of the field is limited (for example, detection of cervical cancer, with 10,000 visual fields of the microscope for each slide). (b) experiments with pre-set time intervals (for example, with intervals of several hours or days). (c) scanning cytometry with preset time intervals, where (a) and (b), when the speed of the autofocus is a fundamental determinant of the temporal resolution.

Whichever source caused the instability, the autofocus will compensate if the positional variations have time constants greater than the autofocus correction.

Generally an autofocus system has an automated microscope, including magnifying lenses and an adjustable stage on which the microscope slide is mounted to achieve an amplified observation of a specimen on the slide. The motors coupled to the stage provide horizontal adjustment of the stage location. A means is provided to carry out the vertical adjustment (Z axis) between the magnifying lenses and the platen. This may include an arrangement for adjusting the vertical position of the objective lens or for the vertical adjustment of the stage. The camera receives an amplified image through the amplifying lens and supplies an electronic signal representing the amplified image for an autofocus electronics. The autofocus electronics processes the signal according to a function that indicates the degree of focus, by providing an adjustment (or error) signal to the vertical adjustment means. In response, the vertical adjustment means adjusts the vertical position of the lens or the stage, changing the focus of the amplified image. Other circuits may be included in an autofocus system for automatic translation (scanning) of a specimen on the slide.

Various methods have been tried for autofocus, including resolution, contrast and entropy. It has recently been shown that the measurement of an optical resolution performs, with validity and precision, the automatic focus. Price, J.H. and Gough DG, "Comparison of Phase Contrast and Automatic Digital Fluorescent Approach for Scanning Microscopy", Cytometry 16, pages 283-297, 1994. This experimental evidence reinforces the following logical definition: the highest resolution always occurs when There is an optimal approach. When an image is out of focus, the details are blurred and the resolution is lost. The resolution can be measured by analyzing the Fourier frequency spectrum with filters that isolate the high frequencies. The sum of the squares of the high frequencies (signal strength) can be used as a resolution measure. In spectral terms, this can be a high pass or bandpass filter. A typical filter consists of the implementation of the first derivative of the intensity of the image. You can also use a Laplacian filter, which is a measure of the second derivative of the intensity of the image. In the Laplacian filter, high pass characteristics predominate, since it measures the resolution on a smaller scale. The transformation into a square signal amplifies the differences between the values of the function.

To compare different criteria, usually an autofocus system calculates the focusing functions as a function of the vertical axis position. The value of the focus function is calculated from an image obtained in each position of the vertical axis. According to Pierce et al., A typical equation of the focusing function using a digital filter is to replicate the ixy image with a high-pass dimensional filter, obtaining the sum of the squares and normalizing to reduce the effect of unstable lighting . That relation is given in equation (1) F (z) = ?? ([- 1 2 -l] * ixy) 2 / [(l / XY of pixels) (?? xy)] 2 (1) where z = vertical position, and ixy is the intensity in the position (x, y).

Analog focus circuits have been reported in Ali Kujoory, M. Mayall, B.H. and Mendelsohn, M.L., "Focus Aid Device for an Exploratory Point Microscope," IEEE Biomedical Engineering Operations, 20 (2), pages 126-32, 1973, and in Johnson, E.T. and Goforth, LJ, "Detection and Focus of Metaphase Extension, Using Closed Circuit Television", Journal of Histochemistry and Cytochemistry, 22 (7), pages 536-587, 1974. McKeogh, L., Sharpe, J., and Johnson, K., in "Low Cost, Automatic Translational and Approach System for Microscope", Meas. Sci. Technol., 6, pages 583-587, 1995, discloses an analog autofocus circuit in microscopy. U.S. Patent No. 5,499,097 describes a microscope autofocus circuit that produces a focus scale signal in which a bandpass filter, a frame circuit, and an integrator are employed. U.S. Patent No. 5,357,280 provides a transient eliminator for removing scanning artifacts in image signals. However, these designs do not take into account, when choosing the high frequency filter, the effect of the transfer function of the autofocus system.

DESCRIPTION OF THE INVENTION The aim of this invention is to implement a fast and accurate, low cost automatic approach, for use, for example, in scanning microscopy of biological specimens colored with fluorescent substances. The invention is implemented in an autofocus system consisting of a microscope, a video signal source representing an amplified image produced by the microscope, and an auto focus control for the microscope which the microscope focuses on in response to a focus scale signal. In this context, the invention is incorporated by an analog circuit that produces the focus scale signal. The circuit includes a filter that provides a filtered signal representing predetermined frequencies in the video signal. A transient eliminator is connected to the filter to eliminate scanning artifacts from the filtered signal. A frame circuit, connected to the transient eliminator, squares the magnitudes of predetermined frequency components of the filtered signal. An integrator connected to the framing circuit produces a focus scale signal representing a degree of focus of the microscope when integrating the square magnitudes of the frequency components of the filtered signal.

Preferably, the circuit further includes an integrator that produces an average illumination signal in response to the video signal. Both the focus scale signal and the medium illumination signal are combined by a processor located in the focusing control of the autofocus system to produce a focus position signal representing a focus position of the microscope. The focus control uses the focus position signal to adjust the microscope to the focus position.

As there is a video signal produced by interlaced scanning, the transient eliminator can also create a window imposed on the video signal to thereby define an area of the amplified image where a focus scale signal will be produced.

Accordingly, one of the objects of the invention is to provide an analog circuit for an autofocus system, which produces a focus scale signal representing a degree of focus.

Still another objective is that such a circuit provides a medium illumination signal, which can be combined with a focusing scale signal, by a processor, to produce a focus position signal.

BRIEF DESCRIPTION OF THE DRAWINGS The objectives, advantages and characteristics of this invention can be more easily appreciated by reference to the following detailed description, when read in conjunction with the attached drawings, in which: Figure 1 is a block diagram of a complete microscope system with autofocus; Figure 2 is a block diagram of an analog circuit for a microscope system with autofocus; Figures 3a and 3b are waveform diagrams illustrating the operation of the analog circuit of Figure 2; Figure 4, constituted by Figures 4A, 4B, 4C and 4D, is a schematic diagram of an electronic circuit that illustrates the best way to implement the analog circuit of Figure 2: Figure 5 is a graph illustrating the digital and analog focus function curves in which the experimental results are shown together with the operation of a microscope system with autofocus with the analog circuit of Figures 2 and 4; Y Figure 6 is a graph showing analog focus function curves at different zooms, with respect to a thick cell monolayer, with the auto focus microscope system of Figure 1.

THE BEST WAY TO PUT THE INVENTION IN PRACTICE Referring now to the Figures in which like reference numerals indicate identical elements, in Figure 1 there is illustrated an automatic focusing system 10 which includes a microscope 12 having an objective 14, and a microscope stage 16 on which You can mount a slide. The plate 16 is translatable in the X and Y directions, so that a succession of areas on the microscope slide, carried on the stage, can be scanned according to known methods. The reference number 18 indicates such an area. The microscope includes a means 20 for illuminating biological specimens marked with a fluorescent substance, and a means 22 for phase contrast or other transmitted microscopy illumination. An image of the area 18 is amplified by magnifying lenses of the microscope 12, which includes an objective 14 and the adjustable focus lens 23. Each amplified image is taken by a camera 30 which produces, by an interlacing scan, a video that represents an amplified image, and includes various synchronization components that are necessary for interlaced scanning. The video signal is supplied in a signal path 32 to a programmed general-purpose digital computer 34, which has, among other functions, an image processor 38 and a focus signal processor 39.

The video signal is also supplied to an analog autofocus circuit 36 embodying the invention. The analog autofocus circuit 36 produces a focus scale signal, a medium light signal, and an activation signal, which are supplied in the paths of the signals 40, 42 and 44, respectively, to the focus processor 39 of the computer 34.

The focus processor 39 of the computer 34 is constituted by a microscope autofocus control that calculates a focus function signal in response to the focus, medium illumination, and activation scale signals. In each of the various focus positions in the microscope 12, a focus function signal is produced. These are combined by the focus processor 39 paxa pioducii a focus position signal representing a focus position of the microscope 12. The focus position signal is supplied in the signal path 50 and is used to control a known medium which it adjusts the position of the objective 14, with which the microscope 12 is focused. In the signal circuit 52, other signals are provided to adjust the X and Y positions of the stage 16.

In column 12, line 28, column 14, line 36 of patent No. 5,548,661 of the United States, an autofocus system is described that does not include the analog autofocus circuit 36, which is fully incorporated by means of this reference.

Hardware and Experimental Methods The autofocus system 10 was adjusted for phase contrast and luminous field illumination, for analog operation according to the invention. Cells were imaged in a Nikon Optiphot microscope through a CF Fluor DL 40x C objective, 0.85 NA with a phase contrast with a Ph3 clarity. Next, the images were further expanded through a Nikon CCTV adjustable focus lens 0.9-2.25 on a Dage VE-1000 CCD RS-170 camera, with a frequency response of 7 MHz. Microscope stage 16 was displaced on the X, Y plane, under the control of computer 34, by means of stepper motors for a fluorescent image cytometry. The control of stage 16 of the microscope was carried out by means of a microgradual driver and an AT ISA computer board compatible with a busbar.

The focus was changed by changing the position of objective 14 with a piezoelectric objective positioner (PIFOC) and a closed loop controller E-S810.10 (Polytec Pl, Costa Mesa, CA). For the displacements of < 1 μm, the position of objective 14 is stable in 10 ms. The position of the PIFOC was controlled by an output from a digital information converter to analog information (D / A) on a data acquisition board DAS 1600 Keithley Metrabyte (Taunton, MA) incorporated into the computer 34. The 12-bit D / A converter divides the 100 μm range of the PIFOC in 4,096 steps of 24 nm each.

The image processor 38, using an RS-170 video input board (the VSI-150 can be obtained from Imaging Technology Inc.), captured the amplified image in the form of a video signal and applied an anti-distortion filter of -3dB to 4.2 MHz and an attenuation of -12 dB to 8 MHz. These values were used as a reference to designate the filters of the analog circuit of this invention and to have a response similar to that of the digital version.

Implementation of the Analog Circuit The invention provides an analog autofocus circuit 36 which measures the degree of focus directly from the video signal of the camera 30 and solves the above design limitations. The block diagram of Figure 2 and the circuit diagram of Figure 4 illustrate the functional components of this novel circuit. The criterion that was used to determine the degree of defocusing was the relative energy contained in the amplified image as a function of the spatial frequency. During blurring, the elements adjacent to the amplified image were blurred or with an average intensity when they were all together, which caused a loss of higher spatial frequencies. By measuring the relative energy of these frequencies as a function of the focal position, the criterion was established to determine the optimal focal position, since the energy changes monotonically and is at maximum when focused. These assumptions are valid for phase contrast only when high frequencies are used for the approach criterion. Under these conditions, monotonicity is often violated when there are low frequencies. Price and collaborators, op. cit.

The analog circuit 36 implements the focus function on the integral of the square values of the video signal, as a measure of the energy of the image. The video signal is filtered before the panning, in order to accentuate the high frequencies that depend a lot on the focus. Equation (2) represents the process achieved by the analog circuit 36 and the computer 34. In terms of the circuit components, the filter 52 selects the range of frequencies from the signal of the video image, and after carrying out framing and integration, the analog circuit 36 produces a focus scale value J j (dlxy / dx) 2dxdy that. it is returned to computer 34 as the magnitude of the focus scale signal together with an average illumination value (Txy) (the magnitude of the average illumination signal). After the A / D conversion, the computer 34, using the focus signal processor 39, squares the average value of the illumination and performs the following division to produce a focus function F (z): F (z) = J í (dlxy / dx) 2 dxdy / (¡Ixydxdy) 2 (2) The shape of the focusing function is determined by the focus criterion, by the microscope and camera transfer functions, and by the object represented. The properties that a useful focus function must have are: 1) unimodality, only a maximum; 2) precision, the maximum occurs in the focus position; 3) reproducibility, the sharpness of the focus function curve; 4) implementation, fast calculation of the focus value. Price and collaborators, op.cit, and Groen, F.C.A., Young, I.T. and Ligthart, G., "Comparison of Different Approach Functions for Use in Autofocus Algorithms", Cytometry 6, pages 81-91, 1985. Analog circuits have a great advantage, over their implementation, over digital circuits, if they can match the operation of the digital circuit in the first three properties. This is because analog components that work with conventional video frequencies can be found almost everywhere and are not very expensive, relatively.

The analog circuit 36 measures the focus directly from the video signal. The output of the camera 30 is in conventional scanned video format and includes two interlaced fields. The analog circuit 36 can be divided into an analog section and a digital / timer section. The analog portion can be further separated into a part of the focusing scale and a part of medium illumination. The analog circuit 36 sends three signals to the computer 34: a focusing scale signal, a medium illumination signal and an activating pulse.

Referring now to Figures 2 and 4, the signal representing an amplified image of the area 18 is sent to a conventional synchronization eliminator 50 which eliminates the horizontal and vertical pulses of the video signal. The output of the synchronization eliminator 50 is fed to the input of a bandpass filter 52 having a transfer function H (?). The filter has a frequency response that passes the high frequency components of the video signal for the reasons discussed above. The filter 52 produces a filtered signal representing the high predetermined frequencies in the image signal. The filtered signal is supplied in an output of the filter 52 which is connected to the input of a transient elimination element 56. The transient elimination element 56 is controlled by synchronization signals extracted from the video signal. In the preferred embodiment, the command eliminates the scanning artifacts produced during the beginning and the termination of each of the scanning lines of the image signal. In fact, the transient elimination element 56 can also be considered as a window generator that, for each scan line having the video signal, allows a window that is smaller than the scan line, with the respective ends of the scan line extending beyond the ends of the window. When the vertical synchronization pulse is supplied, the transient elimination element 56 is allowed to produce a two dimensional window that can be moved over each of the two interlaced fields thus forming a video frame in the typical scanned format. The transient elimination element 56 provides the filtered signal already with the scanning artifacts removed therefrom, in an output that is connected to the input of the quadrant circuit 58. The quadrant circuit squares the magnitude of the predetermined frequency components there are. in the filtered signal, supplying the square magnitudes in an output that is connected to the input of an integrator 60. The integrator 60 integrates the square magnitudes of the frequency components of the filtered signal, thus producing a focus scale signal in the form analog, which is supplied to a sampling and holding circuit 62. The sampling and holding circuit 62 is commanded to retain a voltage magnitude of the integrated signal produced by the integrator 60. The voltage magnitude of the integrated signal (the signal of focus scale) represents a degree of focus of the microscope 12. The focus scale signal is supplied in the path of the signal 42 to the computer 34. Consequently, the elements 52, 56, 58, 60 and 62 form a part of the focusing scale of the analog portion of the analog circuit 36. A part of the average illumination of the analog circuit 36 is formed by an integrator 68 that integrates the video signal, its synchronization signals having been eliminated by the synchronization eliminator. The integration of the video signal by the integrator 68, for example on a video line, represents the average illumination of the entire line. The magnitude of the integrator 68 is sampled and retained by the sampling and holding circuit 70 whose output forms the average illumination signal supplied in the path of the signal 42.

The digital portion of the analog circuit 36 is constituted by a control timer circuit 66 which receives intact the signal of the image, including all its scanning artifacts, such as the vertical and horizontal synchronization portions. The control timer circuit 66 generates reset and hold signals which sequentially synchronize the operations of the integrators 60 and 68 in the sampling and holding circuits 62 and 70 respectively. In addition, the control timer circuit 66 produces the activation signal on the signal line 46.

The control timer circuit 66 also produces the synchronization signals necessary to form the window implemented by the transient eliminating element 56.

Digital Section With reference to Figures 2 and 4, the synchronization pulses of the video signal are detected in the control timer circuit 66 by a synchronization separator 80 (LM1881, National Semiconductor, Arlington, TX) which extracts the horizontal pulses and vertical This timing information is used to create a window that represents an area of the image where the focus function will be implemented. The least that the window will do is eliminate the discontinuities generated by the filter 52 at the ends of the horizontal lines. This portion of the analog circuit 36 can be used as a masking generator to select an arbitrary rectangular portion of a video field for processing; any window size can be defined vertically and horizontally by changing the time constants of the two monostable multivibrators. The biostable circuits 82a and 82b establish a first portion of the window; the second portion is established by the biostable circuits 83a and 83b. This type of analog masking has been used for video-dimension analyzers. Yin, F.C.P., Tompkins, .R., Peterson, K.L. and Intaglietta M, "Video-Dimension Analyzer" IEEE Operations of lnyenieiid Biomedical, 19 (5), pages 376-81, 1972. The information of the window is used by a controlled amplifier 84 located in the transient elimination element 56 After each window, or after each video field, an activation pulse is generated. The control timer circuit 66 also produces a sequence of 60 Hz trigger signals that command the computer 34 for an A / D conversion of the corresponding analog values of each field.

Analog Section The filter 52 has built-in 90-96 wideband monolithic amplifiers that have high response speed and internal compensation of unit gain frequency for high speed and stability. Such high frequency and high-speed amplifiers tend to have more oscillations than low frequency devices. However, this instability was eliminated by reducing the parasitic capacitance on the inputs and outputs of the amplifiers. The derivation of the power supply was also used to improve the instability, and small capacitors were added in parallel to the feedback resistors to compensate for parasitic capacitance not preventable in the filters.

Figure 3a shows the graph of a horizontal video line 100 and the subsequent selected analog processed outputs. The synchronization eliminator 50 removes the synchronization portion of the composite video signal for each waveform 102.

The reference level is in the mass, and since the tip of the synchronizer is negative, the output will have eliminated the synchronizer and will place the suppression level in the mass. After the synchronization pulses are eliminated at the input of the video signal, the signal is fed both to the filter 52, in the focus scale section, and to the lighting integrator 68.

Each of the operational amplifiers 90-96 of the filter (LT1220, Linear Technology Corporation, Milpitas, CA) are used in a Butter orth bipolar active filter configuration, with the four operational amplifiers arranged in a low pass section 52a of the fourth order and in a high pass section 52b of the fourth order. A frequency response of 2 to 4 MHz was selected to be equal to the response of the digital filter, which represents the transfer function of the lenses in the microscope 12. A bandpass gain of 2.56 was used to compensate the attenuation of the signal. Due to the wide bandwidth and the unit compensation of the components, good operation was obtained without having to make any other special modification. By choosing the components well, the cutoff frequency is independent of the bandwidth of the amplifier, and is determined only by the respective R-C (resistance-capacitance) networks in the low pass section 52a and the high pass section 52b. Consequently, these networks can include manually adjustable elements. Waveform 104 shows the filter output.

The filtered and windowed signal is compensated, amplified and squared, as shown in waveforms 106 and 108. Then, the filtered and square signal is integrated into a video field by the integrator 60. The integrator 60 performs the restart, integration, and hold control functions to reset the capacitor 110 at the end of each field, integrate the filtered signal to calculate the focus scale, and, at diode 114, to maintain an intermediate focus while there is no output important of the filter 52. The previous designs of automatic circuits of autofocus, which lacked the control of retention, showed a decrease in the output of the scale of focus between the characteristics of the image. This is clearly illustrated in the waveform 109 of Figure 3a, where, on the portion of the square waveform 108 marked "dead space", the magnitude of the integrated signal produced by the integrator 60 is not decremented. Using the integrator 60 with the diode 114, instead of the conventional integrator, it is ensured that the focus scale equals more ideally the true mathematical integral of each video field. The output of the integrator 68 is controlled in a similar manner by the capacitor 116 and the diode 118.

The middle lighting section takes the video signal with the timing removed and measures the average lighting by integrating the signal in a field. A sampling and holding circuit is used to maintain the final analog values of the A / D board. The final output of this integration, and the focus scale, are converted into computer 34 by the analog-digital board. The activation signal is sent to the computer at the end of each field, to start a new conversion. The focusing scale, the average illumination and the activator, with their respective masses, are connected in a conventional manner to the computer 34 or to the paths of the signals 40, 42 and 46, respectively.

The selection of analog circuit windows 36 is illustrated in Figure 3b, and can be understood with reference to Figures 2 and 4. For the dimension of the scan line, a horizontal window pulse 120 is created, for any line of video, by the bistable circuits 82a-82b, in response to the suppression of horizontal line. A vertical window pulse 124 is produced by flip flops 38a-38b in response to vertical blanking. The transistor circuit-diode 126 responds to the pulses 120 and 124 as gate Y, transmitting the filtered signal from the filter 52 to the square circuit 58 when both pulses are high. Obviously, counting or timing can be employed with the bistable circuits of the control timing circuit 66 to selectively adjust the dimensions of a window.

Focus Software, Automatic Approach Algorithm In the focus processor 39, an interrupt processing routine (TSR) was incorporated to obtain the analog value of the focus and mid-illumination scale signals, based on the trigger signal sent by the analog circuit 36. This routine also controlled the position of the focus, obtaining analog values from the autofocus circuit and calculating the normalized degree of focus. The programs were written in C languages and assembler language (assembly language). The routines in C were compiled with Metaware High C (Santa Cruz, CA). A Phar Lap assembler (Cambridge, MA) was used for the interruption treatment routines.

At the end of each field, an activation pulse initiates the interrupt processing routine, which transfers the analog values of the focus and mid-illumination scale signals to sets of values accessible to the C routines, in order to calculate and Establish the best approach.

In order to establish the best approach, the computer 34 executes a focusing sequence in which the Z-axis position (vertical axis) of the objective 14 is sequenced by a variety of focus positions (zi). At each focus position, the magnitude of the focus and mid-illumination scale signals is taken, and a focus function value (F (z?)) Is calculated according to equation (2). The values of the focus function are stored in 130, Figure 1, by the computer 34.

After each focusing sequence, with the focus scales stored by the computer 34 for various positions, a weighted average is used to search for the best focus. Focus curves with unusual shape, containing multiple ends, can be produced by discrete vertical distributions of cellular components. For these reasons, the weighted average is used Wa = S z (Fz) "(3) S (FZ)" , where a is the weighted average position, z is the vertical position (Z-axis), Fz is the result of the focus function (equation (2)) calculated from an image obtained at a position z, and n is the power of the weighting. The power accentuates the peak values and the average reduces the effect of the three-dimensional (3D) nature of the specimen.

Experimental Results The contrast in a microscope image is not an inherent property of a specimen. Rather, it is a product of (1) the interaction of luminous light waves and the structure of the specimen and (2) the mode of generation of contrast and MTF of the microscope. Point (1) depends on both the structure of the specimen and the condition of the wave of light; point (2) depends both on the condition of the illumination and on the treatment given to the waves coming out of the specimen. Inoué, S., "Video Microscopy" (Video-Microscopy), Plenum Press, New York, 1906.

In these experiments, phase contrast microscopy was used as a technique for reproducing images. In the phase contrast, the changes introduced by the transparent cells are transformed into intensity changes. Born, M. and Wolf, E., "Principles of Optics", Pergamon Press, 1989. This creates a contrast in the image, which is useful for performing autofocus. The phase contrast also constitutes a high-pass optical filter. Inoué, S., "Video Microscopy", Plenum Press, New York, 1986. It has also been demonstrated experimentally that phase contrast has a greater tendency to present lateral peaks in the focus function curve . Price, J.H. and collaborators, op. cit. Consequently, it is important to select high frequencies very carefully, to ensure unimodality.

The graph of a phase contrast experiment of a cell monolayer in different Z-axis positions is illustrated in Figure 5. With similar analog and digital versions, similar focus function curves are obtained; the width of the peaks and the sharpness of the functions are mainly unimodal. Outside the main lobe, the curves have cushioned lateral peaks. This behavior is exacerbated if there is not enough sampling in the unit variable optics. Increasing the amplification results in Nyquist sampling and this behavior is eliminated. The low cutoff frequency of the filter captures mid-range frequencies that can not be assumed with a monotonic behavior. The tendency towards lateral peaks is reduced with the increase in the frequency response of the transfer function of the focusing scale system, which includes the filter, the CCD camera and the optical transfer function.

Figure 6 shows a graph of the analog focus scale using a thick cell monolayer. The three-dimensional structure of the specimens creates differences regarding the best approach and is responsible for expanding the focus function. It seems that a greater depth of the specimen improves the lateral peaks. Consequently, the shape of the focus curve depends on both the specimen and the transfer function of the system. The sampling period for a particular experimental condition was calculated using images from a micrometer slide with a separation of 10 μm. With a zoom of lx (using a 40x objective) the period was 303 nm. The resolving power is provided by the Rayleigh criterion, d = l .22? (NAobj + N c? Nd). With an illumination of 500nm, 0.52 N c? Nd and 0.85 NAo, a resolution of 445nm is obtained. Consequently, the required Nyquist sampling is 222.5nm. Consequently, an amplification of (303 / 222.5) = 1.36x is required for Nyquist sampling. Due to various practical reasons, in fact more tests must be done. Inoué, S. , op.cit.

The main advantage of this carefully designed analog circuit is that it has a low cost without sacrificing performance. Digital processing for real-time autofocus requires the use of a real-time image processor, with a pipe architecture that can be an order of magnitude more expensive. In addition, to increase the complexity of the digital filter it is necessary to add more coefficients, which could increase the cost or reduce the speed. On the other hand, digital autofocus makes reprogramming a simple task; and for real-time operation, digital image processing resources can be used for simple one-dimensional filters, such as discrete approximations of derivative filters (for example, high-pass filters {1, -1.}. { -1.2, -1.}., And the bandpass filter { 1,0, -1.}.). However, due to the cost of digital image processing, it should be taken into account to replace the digital implementation with an analog circuit of low cost and equal performance, in order to leave this important resource free and apply it to other tasks.

In general, a disadvantage of analog circuits (at least with respect to its digital equivalents) is its limited dynamic range. The digital dynamic range is proportional to the square root of the product of the pixel and gray levels. Analog processing is limited to the number of bits in an A / D converter. Of course, the limited analog dynamic range can be overcome by adding an automatic gain control circuit. On the other hand, the analog implementation allows an arbitrary upper cutoff frequency (up to the camera boundary), while in the digital version this upper limit is set by the image processor. This makes it easier to match the focus cutoff frequency, existing in the filter 52, to the optical transfer function of the microscope, so as to generate a sharper filter function curve and achieve improved auto-focus reproducibility. Assuming that the image processor 38, with a CCD camera capable of creating 768 pixels per line, digitize only 512 pixels per line. With the analog circuit 36, plug-in heads for the filter 52 can be used and thus simplify the equalization of the filter function with each combination of video camera and optical transfer function. The analog circuit filter 52 is much easier and less expensive to change than the resolution of the image processor 38.

Although the previous detailed description has illustrated, described and pointed out the fundamental novel features of the invention, as they apply to different embodiments, it should be understood that those who have knowledge of the subject will be able to carry out different omissions, substitutions and changes to the form and details of the illustrated device, without departing from the spirit and scope of the claimed invention.

Claims (12)

1. A circuit (36) for producing a focus scale signal in an autofocus system having a microscope (12), a source (30) of a scanned image signal representing an amplified image produced by the microscope, and a automatic focusing control (34) for the microscope, which focuses the microscope in response to the focus scale signal, said circuit (36) including a filter (52) for supplying a filtered signal, which represents certain frequencies in the signal of the image, a transient eliminator (56) connected to the filter in order to eliminate the scanning artifacts from the filtered signal, a square circuit (58) connected to the transient eliminator to square magnitudes of the frequency components of the filtered signal, and a scale integrator (60) connected to the square circuit to produce a focus scale signal representing the microscope focus by integrated square magnitudes of the frequency components of the filtered signal, the circuit (36) being characterized by the following: The filter is an analog filter that has a transfer characteristic (H (?)) That only passes the upper half of the frequencies in the optical transfer characteristic of the microscope.

2. The circuit of claim 1, further including a diode (114) at the output of the scale integrator, to prevent the focus scale signal from decreasing.

3. The circuit of claim 1, wherein the transient eliminator applies a one-dimensional window to the filter signal.

4. The circuit of claim 1, wherein the transient eliminator applies a two dimensional window to the filter signal.

5. The autofocus control system of claim 1, wherein the focus scale signal indicates the resolution of the amplified image.

6. The circuit of claim 1, wherein the signal of the image is a video signal (100).

7. The circuit of claim 6, wherein the scanning artifacts include the ends of a scan line of the video signal.

8. The auto focus control system of claim 6, wherein the focus scale signal indicates the energy that is contained in predetermined frequencies of the video signal.

9. The circuit of claim 1, further including an illumination integrator (68) for producing a medium illumination signal in response to the image signal.

10. The circuit of claim 9, wherein the focus control includes a processor (39) for producing a focus position signal representing a focusing function of the microscope, in response to the focus scale signal and the signal of the average lighting.

11. The circuit of claim 10, which also includes: a first diode (114) at the output of the scale integrator, to avoid decreasing the focus scale signal; Y a second diode (118) at the output of the illumination integrator, to prevent the average illumination signal from decreasing.

12. The autofocus control system of claim 11, wherein the processor produces a focus function value for each of the various focus positions of the microscope, by combining a focus scale signal value and a value of the average illumination signal that is obtained at each focus position, and produces a focus position signal representing the focus position of the microscope, by combining a diversity of focus function values.