METHOD FOR OPERATING A CMOS IMAGE SENSOR AND CMOS IMAGE SENSOR AND ITS USE

Abstract

The present invention relates to a method for the fast and energy-efficient operation of a CMOS image sensor (1), which can be used in particular in laser triangulation or similar methods that project a light line with a Gaussian intensity profile onto the CMOS image sensor (1). The method according to the invention is characterized in particular by the fact that the brightness (Hiv;ih) of the individual pixels (2) is determined solely on the basis of time markers (ti) stored in a memory (33), thus preventing overexposure of the pixels (2) and consequently of the CMOS image sensor (1), and thereby increasing its dynamic range. Furthermore, only pixel information (time marker (ti) and position (iv;ih)) from pixels (2) whose exposure has reached a predetermined threshold (T) is processed, and unexposed pixels (2) are disregarded.Since the power consumption of an image sensor is proportional to the data throughput, the data processing and transmission optimized according to the invention not only lead to an increase in the scan rate, but also to lower power consumption and thus to a particularly energy-efficient operation of the CMOS image sensor (1).

Description

[0001] The present invention relates to a method for operating a CMOS (complementary metal-oxide-semiconductor) image sensor, which can be used in particular in object measurement and / or object inspection methods in which only a part of the pixels of a CMOS image sensor is exposed.

[0002] In a number of object measurement and / or object inspection methods, one or more individual light points or one or more light lines are detected using a CMOS image sensor, so that only a small part of the pixels of the CMOS image sensor is exposed and the rest of the pixels remain unexposed.

[0003] In the field of 3D object or surface inspection and / or surveying, for example, laser triangulation involves projecting a laser light point or laser line onto a measurement object from one direction and observing the reflected light from another direction with a sensor. If the distance between the laser generator and the sensor, as well as the angle between the laser beam and the observation direction, is known, the distance to the measurement object can be determined. Within scanning methods based on this laser triangulation, the laser beam is moved across the object under investigation, thereby detecting one or more reflection points or lines moving across the CMOS image sensor. Other examples in this field include optical coherence methods and so-called...“Flying spot applications” involve moving a single laser point across a surface under investigation using a galvanic laser or a MEMS (microelectronic-mechanical systems) laser. In the 2D domain, an example would be X-ray single-crystal diffractometry, in which monochromatic X-rays are scattered by the atoms of a crystal lattice and detected by a CMOS image sensor in the form of so-called “single-crystal diffractograms”.

[0004] Conventional image sensors are only conditionally suitable for use in such processes, where a large proportion of the image sensor's pixels remain unexposed.

[0005] For example, if the reflected light from a laser line reflected off a measuring object is imaged onto a sensor matrix of pixels arranged regularly in rows and columns, the laser line often lies primarily horizontally in the image and covers only a small number of pixels per column (e.g., 10 pixels / column). Assuming that only the brightest four of these 10 pixels are truly relevant, a column length of 1000 pixels / column results in a coverage of only 0.4% of the entire image area. The majority of pixels in a column thus remain unexposed or contribute nothing relevant to the image information, yet are still read out to a disadvantage by conventional image sensors.Since the number of pixels to be read determines the frame rate and thus the scan rate of a sensor used in laser triangulation or similar methods, conventional image sensors therefore slow down the scanning process and are inefficient.

[0006] One way to address this problem with CMOS image sensors is to limit the readout area, both in terms of rows and columns, to smaller regions—a process known as windowing. This improves the ratio of relevant pixels read to the total number of pixels read, but it still results in a majority of pixels with little to no information being read. Furthermore, limiting the readout area also reduces the image area (measurement range).

[0007] Another problem with using conventional image sensors in laser triangulation or similar methods lies in the sometimes very large differences in brightness that need to be detected. Matte white surfaces of measurement objects are significantly brighter than black or glossy surfaces. Differences of 100:1 or more in the intensity of reflected light are not uncommon. The dynamic range of conventional image sensors is far from sufficient to cover these large differences in brightness. While established operating methods for image sensors, such as multiple exposure, offer a good compromise, they negatively slow down the sensor's scan rate. Using amplifiers with logarithmic or multiple-slope characteristics compromises accuracy because these approaches shift the center of brightness.When using amplifiers with a linear characteristic curve, at least 16-bit ADCs (analog-to-digital converters) would be required to achieve sufficient dynamic range. As a result, both the hardware complexity and power consumption would increase drastically.

[0008] Based on this, the present invention aims to provide a method for operating a CMOS image sensor, or a CMOS image sensor and its use, in which no saturation occurs and whose dynamic range, i.e., the ratio between the maximum and minimum measurable number of electrons of the saturation capacitance, is practically arbitrarily high, preferably at least 100 dB. Furthermore, the scan rate should be as high as possible, while simultaneously avoiding any limitation of the image or measurement range.

[0009] This problem is first solved by a method for operating a CMOS image sensor with the features of independent claim 1 and by a CMOS image sensor with the features of dependent claim 12 and its use according to claim 15.

[0010] A CMOS image sensor according to the invention comprises at least a regular arrangement of pixels in columns and rows; wherein the pixels each comprise at least one photodetector and at least one comparator; and a control unit, wherein the control unit comprises a counting unit for processing a clock signal and for generating time markers, and at least one memory; and wherein the control unit is configured to specify at least one threshold for a photon incidence.

[0011] A method according to the invention for operating such a CMOS image sensor comprises at least the following method steps: - Resetting the pixel voltage of the photodetectors of all pixels to a defined initial value; - Reset all timestamps to zero; - Clearing the memory; - Starting the clock signal when an exposure of the CMOS image sensor begins; - Detecting each photon incidence using the photodetector of the respective pixel; - Comparing the actual value of the detected photon incidence of each pixel with the threshold value for photon incidence specified by the control unit using the respective comparator of the respective pixel; - if the threshold value for a pixel is exceeded, the counting unit generates a timestamp for that pixel; - Storing the timestamp for this pixel in memory; - Reading the memory; and - Determining the brightness of individual pixels based on the timestamps stored in memory.

[0012] The method according to the invention advantageously processes only pixel information (time stamp and, if applicable, position) of pixels whose exposure has reached a predetermined threshold. Unexposed pixels are disregarded. Since only relevant pixels are read out, the method according to the invention is characterized by very high efficiency and a very high frame rate or scan rate, and thus high speed. In contrast to methods for increasing efficiency, such as prior art windowing, which also do not read out and process all pixels of an image sensor, the method according to the invention advantageously allows the entire image area of ​​the CMOS image sensor to be used. Selection of the readout area by windowing, and thus the corresponding configuration effort, is eliminated.

[0013] Furthermore, the inventive method advantageously determines the brightness of the individual pixels solely on the basis of the time stamps stored in memory. The inventive method advantageously utilizes the fact that the brightness of a pixel is inversely proportional to its assigned time stamp: the higher the photon incidence on a particular pixel, the faster a predetermined threshold is exceeded, and the smaller the counter value that a counting unit assigns to the respective pixel as a time stamp. The individual pixel signals, and in particular the pixel voltages stored in integrating capacitors, are used only for comparison with the aforementioned threshold value and are not processed further. The threshold value can be chosen significantly below the saturation value of the pixels, thus advantageously avoiding overexposure of the pixels and consequently of the image sensor. Furthermore, for evaluation or...Further processing of pixel data advantageously does not require waiting for a further clock signal or the end of the exposure of the entire image sensor. The moment a timestamp is assigned to a pixel, this value is available for further processing. CMOS image sensors operated using the method according to the invention thus advantageously exhibit very short latency times.

[0014] The method according to the invention advantageously improves the data transmission from the pixel field of the CMOS image sensor by considering only relevant pixels. Since the power consumption of an image sensor is proportional to the data throughput, less data advantageously leads to lower power consumption and thus to particularly energy-efficient operation.

[0015] Further advantageous designs and advanced features, which can be used individually or in combination, are the subject of the respective dependent claims.

[0016] In a preferred embodiment of the invention, it has proven advantageous if the memory is at least a columnar memory, in particular a columnar memory operated according to the FIFO ("first in, first out") principle, and the timestamp generated for a specific pixel is stored in the columnar memory together with a vertical position of the pixel, in particular with the respective row number of the CMOS image sensor of the respective pixel. With this storage variant, the horizontal position, in particular the column coordinate, of the pixel advantageously does not need to be stored, which saves memory space and speeds up the storage process. In an embodiment of the invention in which, for example, two columnar memories are provided, these can advantageously be operated alternately.For example, one of the two column memories can be written to while the other is being read, thus advantageously increasing the sampling rate of the CMOS image sensor. Alternatively or cumulatively, the memory can also be at least one row memory, in particular a row memory operated according to the FIFO ("first in, first out") principle, and the timestamp generated for a specific pixel, together with a horizontal position of the pixel, in particular with the respective column number of the CMOS image sensor of the respective pixel, can be stored in the row memory.

[0017] Alternatively, an embodiment of the inventive method has proven successful in which the memory is at least a global memory, in particular a global memory operated according to the FIFO ("first in, first out") principle, and in which the timestamp generated for a specific pixel is stored in the global memory together with the horizontal and vertical position of the respective pixel on the CMOS image sensor, in particular together with the respective column and row number of the CMOS image sensor of the respective pixel. Before reading the memory, the stored value triples comprising the timestamp and the associated horizontal and vertical position of the respective pixel can be sorted by column or row. After the end of the image generation or...With this storage method, all data records are advantageously pre-sorted by columns or rows before further processing during the scanning process.

[0018] In a further preferred embodiment, it has proven advantageous if the control unit specifies two or more threshold values ​​and each pixel correspondingly comprises two or more comparators, in particular one comparator for each threshold value specified by the control unit. The ability to specify at least one threshold value via the control unit advantageously allows, among other things, adjustment to the laser intensity, exposure time, and the properties of the object surface under investigation (reflectivity). If each pixel comprises two or more comparators, the respective stored pixel voltage can advantageously be compared with two or more threshold values ​​specified by the control unit, thereby generating and assigning a time marker earlier, particularly for darker pixels.Such a design also makes it possible, in particular, to provide a higher threshold for the brightness maximum and lower thresholds for the secondary values, thereby advantageously reducing motion blur and improving the simultaneity of brightness detection. Furthermore, such a design with two or more thresholds specified by the control unit and correspondingly two or more comparators per pixel advantageously enables the measurement of the laser intensity gradient, which, among other things, advantageously allows for compensation of the pressure level of the CMOS image sensor.

[0019] Furthermore, it is advantageous if the generation and storage of timestamps for pixels within a column is aborted after a predetermined number of pixels, particularly after 2, 3, 4, 5, 6, or more pixels, wherein the predetermined number of pixels is particularly preferably even, especially 2, 4, or 6 pixels. Aborting the generation and storage of timestamps for pixels within a column after a predetermined number of pixels advantageously increases the evaluation speed of the CMOS image sensor and saves memory space. In addition, power consumption is advantageously reduced. This is particularly beneficial when using the CMOS image sensors according to the invention in laser triangulation for measuring 3D objects (machine parts, etc.).In 3D object acquisition, this approach can be advantageously used to increase efficiency, since the laser line focused on the object under investigation typically covers only a small number of pixels within a column—usually three pixels—and exhibits a Gaussian intensity distribution. Terminating the laser line after, for example, five pixels within a column is therefore sufficient. Other pixels within the respective column can be advantageously disregarded. If the number of predetermined pixels is preferably chosen to be even, the efficiency of the subsequent data analysis is advantageously increased, particularly the determination of a center of gravity (COG) within the respective column. In principle, two pixels can suffice for such an analysis, although the addition of two or four secondary pixels can advantageously increase the accuracy of the COG determination.

[0020] In a further preferred embodiment of the invention, the generation and storage of timestamps for pixels by the control unit can be permitted only for a predefined area within a column, in particular only for predefined rows within a column. In this way, the readout area of ​​the CMOS image sensor, similar to the multi-windowing in prior art image sensors, can advantageously be limited from the outset to smaller column sections that are of interest for evaluation, thereby advantageously suppressing stray light from reflections.

[0021] It is also advantageous if, in one process step, the brightness of the individual pixels of a column is determined - first, the pixel with the smallest time stamp is determined and classified as the pixel with the greatest brightness; - then the pixels adjacent to the pixel with the greatest brightness within a column are compared with each other, and the pixel with the next largest time stamp compared to the smallest time stamp is assigned the next lowest brightness, and so on. - until a brightness is assigned to all pixels, especially all pixels for which timestamps are stored in memory.

[0022] To determine the brightness of the individual pixels in a row, the following can be used: - first, the pixel with the smallest time stamp is determined and classified as the pixel with the greatest brightness; - then the pixels adjacent to the pixel with the greatest brightness within a row are compared with each other, and the pixel with the next largest time stamp compared to the smallest time stamp is assigned the next lowest brightness, and so on. - until a brightness is assigned to all pixels, especially all pixels for which timestamps are stored in memory.

[0023] This approach advantageously determines the brightness of individual pixels based on previously stored timestamps or their sequence. It eliminates the need to detect the absolute number of photons hitting each pixel, thus avoiding pixel saturation. Optionally, the pixel classified as having the highest brightness—the "brightest pixel"—can be normalized to 100% brightness to reduce the complexity of the computational logic.

[0024] In a further preferred process step, a distribution of the brightness of the individual pixels can then be determined; - within a column: - the respective brightness of each pixel marked with a time stamp is multiplied by the vertical position of the respective pixel; - a sum of all products obtained in this way from brightness and vertical position is formed; - the obtained sum is divided by the sum of all brightness values ​​of the pixels in the respective column that are marked with a time stamp; - and thereby a subpixel-accurate vertical position of a center of brightness within the respective column can be determined.

[0025] Alternatively or cumulatively, - within a single line: - the respective brightness of each pixel marked with a time stamp is multiplied by the horizontal position of the respective pixel; - a sum of all products obtained in this way from brightness and horizontal position is formed; - the obtained sum is divided by the sum of all brightness levels within the respective row; - and thereby a subpixel-accurate horizontal position of a center of brightness within the respective line can be determined.

[0026] Furthermore, in another preferred embodiment of the method according to the invention, the path of a laser line across the regular arrangement of pixels in columns and rows of the CMOS image sensor can be determined by determining the vertical position of the centroid of brightness for several columns, preferably for each individual column. If the vertical position of the centroid of brightness is determined for several columns, the path of the laser line can advantageously be obtained by connecting said vertical positions. Determining the vertical position of the centroid of brightness for each individual column advantageously enables the greatest accuracy.

[0027] Finally, a specific embodiment of the method has proven effective in which the exposure of the respective pixel is interrupted upon reaching the threshold value, in particular by opening a photocurrent switch assigned to the respective pixel, wherein preferably an integrating capacitor assigned to the respective pixel is discharged by means of a reset circuit assigned to the respective pixel. Such a procedure advantageously reduces electrical crosstalk, i.e., unwanted mutual interference of neighboring pixels by electrical fields.

[0028] The control unit of the CMOS image sensor is configured according to the invention to carry out the method described above.

[0029] In a preferred embodiment of the CMOS image sensor according to the invention, each pixel comprises two or more comparators.

[0030] In a particularly preferred embodiment of the CMOS image sensor according to the invention, the CMOS image sensor can be configured as a BSI sensor (back-side illumination, BSI). CMOS image sensors designed as BSI sensors, i.e., as semiconductor image sensors with back-side illumination, advantageously exhibit higher light sensitivity and improved image quality.

[0031] The CMOS image sensor according to the invention is particularly well suited for use in object measurement and / or object inspection methods in which only a portion of the pixels of a CMOS image sensor are exposed. Such methods include, for example, those in which a line of light with a Gaussian intensity profile is imaged onto the CMOS image sensor, such as laser triangulation; or those in which a single light point or several individual light points are imaged onto the CMOS image sensor, as in so-called "flying spot applications" or diffractometry methods.

[0032] Additional details and further advantages of the invention are described below with reference to preferred embodiments, to which the present invention is not limited, and in conjunction with the accompanying drawing.

[0033] This schematically illustrates: Fig. 1. An example of a general structure of a CMOS image sensor; Fig. 2. An embodiment of a CMOS image sensor according to the invention with two pixels in the form of a block diagram; and Fig. 3. By way of example, the course of a laser line over a section of the image area of ​​a CMOS image sensor according to the invention, as well as in an enlarged representation a section of the first column and the associated intensity profile of the laser line in said section.

[0034] In the following description of preferred embodiments of the present invention, the same reference numerals denote identical or comparable components.

[0035] Fig. Figure 1 shows an example of a general structure of a CMOS image sensor 1.

[0036] CMOS image sensors in general, as well as the CMOS image sensor 1 according to the invention, comprise at least a regular arrangement of pixels 2 in columns S1, S2, ..., S x and lines Z1, Z2, ..., Z x , which form their image area, and a control unit 3 for controlling and evaluating the image data. Data exchange between the control unit 3 and the individual pixels 2 can take place directly and / or indirectly, in particular via a row access circuit 11 and / or a column access circuit 12.

[0037] In Fig. Figure 2 shows an embodiment of a CMOS image sensor 1 according to the invention with two pixels 2 as an example, in the form of a block diagram.

[0038] According to the invention, each pixel 2 comprises at least one photodetector 21 for detecting an incident photon and at least one comparator 25. The control unit 3 according to the invention comprises a counting unit 31 for processing a clock signal 32 and for generating time markers t. i as well as at least one memory 33. Said memory 33 can preferably be operated according to the FIFO (first in - first out) principle. The control unit 3 is furthermore configured to specify at least one threshold value T for the photon incidence.

[0039] According to the invention, such a CMOS image sensor 1 is now operated as follows: Before a new scan or measurement is started with the CMOS image sensor 1, the pixel voltages of the photodetectors 21 of all pixels 2 are first reset to a defined initial value, in particular to zero, as well as any stored time stamps t. iThe value is set to zero. Additionally, memory 33 is erased. This can preferably be done via reset circuits R of pixel 2, counter unit 31, and memory 33.

[0040] When the CMOS image sensor 1 is exposed, in particular by one or more, especially two or three, laser lines L during a laser triangulation measurement, the clock signal 32, generated, for example, by a reference clock generator, is started simultaneously. When light (photons) falls on the photodetector 21 of a pixel 2, the photocurrent 22 generated in the photodetector 21 can flow via a closed photocurrent switch 23 to an integrating capacitor 24, where it is stored as a pixel voltage, thus detecting the photon incidence on the respective pixel 2. The comparator 25 of the respective pixel 2 compares the actual value of the detected photon incidence with the threshold value T for photon incidence specified by the control unit 3, whereby the control unit 3 can transmit said threshold value T to the comparators 25 of the individual pixels 2 in the form of a comparison voltage.If the actual value of the photon incidence of a pixel 2, i.e., in particular its pixel voltage, exceeds the threshold value T, i.e., in particular the value of the specified reference voltage, a time marker t is set by the counting unit 31. i The value for this pixel 2 is generated and stored in memory 33. The counter unit 31 preferably functions as a global counter, which processes the clock signal 32 such that whenever the threshold value T is exceeded at a pixel 2, a copy of the counter value of the counter unit 31 at the time of the exceedance is generated as a timestamp t. i The value for this pixel 2 is stored in memory 33. If the pixels 2 include two or more comparators 25, the control unit 3 can also specify two or more threshold values ​​T, so that darker pixels 2 can be advantageously detected earlier.

[0041] The timestamp t generated for a specific pixel 2 ican be in a memory designed as a columnar memory 33 together with the vertical position i v of pixel 2, in particular with the respective number of the row Z1, Z2, ..., Z x of the CMOS image sensor 1 of the respective pixel 2, and / or in a memory 33 designed as a line memory together with the horizontal position i h of pixel 2, in particular with the respective number of column S1, S2, ..., S x of the CMOS image sensor 1 of the respective pixel 2, are stored. The time marker t i but can also be used together with the horizontal i h and the vertical i v Position of the respective pixel 2 on the CMOS image sensor 1, in particular together with the respective number of column S1, S2, ..., S x and the respective number of line Z1, Z2, ..., Z xThe CMOS image sensor 1 of the respective pixel 2 is stored in a memory 33 designed as a global memory. This storage variant is advantageous, for example, for the detection of point-like events, especially in the context of flying spot applications.

[0042] In the case of a memory 33 designed as a global memory, the stored value triples comprising a time stamp t can preferably be read out even before the memory 33 is read. i and associated vertical i v and horizontal i h Position of the respective pixel 2 according to columns S1, S2, ..., S x or lines Z1, Z2, ..., Z x to be sorted.

[0043] To further increase readout efficiency and speed, timestamps can be generated and stored. i for pixel 2 within a column S1, S2, ..., S xThe process can be terminated after a predetermined number of pixels 2, in particular after 2, 3, 4, 5, 6, or more pixels 2, wherein the predetermined number of pixels 2 is preferably even, in particular 2, 4, or 6 pixels 2. Alternatively or cumulatively, the generation and storage of timestamps t can be performed. i For pixel 2 from control unit 3, only for a previously defined area within a column S1, S2, ..., S x , especially only for previously defined rows Z1, Z2, ..., Z x within a column S1, S2, ..., S x , be admitted.

[0044] After reading the memory 33, preferably by means of a data readout unit 34, the brightness H is then determined according to the invention. iv;ih of the individual pixels 2 based on the timestamps stored in memory 33 i determined. To determine the brightness H iv;ih of the individual pixel 2 of a column S1, S2, ..., S xFirst, especially within a computing unit 35, the pixel 2 with the smallest time stamp t can be selected. min determined and designated as pixel 2 with the highest brightness H max They can be classified. Then, those corresponding to Pixel 2 with the highest brightness H can be identified. max adjacent pixel 2 within a column S1, S2, ..., S x Each will be compared with each other, and the Pixel 2 will have the smallest time stamp compared to the smallest. min next largest time marker t min+1 the next lower brightness H max-1 , are assigned, and so on. This assignment of relative brightness values ​​H max , H max-1 , H max-2 , ...preferably continues until all pixels 2, in particular all pixels 2 for the 33 timestamps stored in memory, have been updated. i are stored, a brightness H i is assigned. To determine brightness H iv;ih of the individual pixels 2 of a row Z1, Z2, ..., Z xAccordingly, the pixel 2 with the smallest time stamp t can first be selected, especially within a computing unit 35. min determined and designated as pixel 2 with the highest brightness H max They can be classified. Then, those corresponding to Pixel 2 with the highest brightness H can be identified. max adjacent pixel 2 within a row Z1, Z2, ..., Z x Each will be compared with each other, and the Pixel 2 will have the smallest time stamp compared to the smallest. min next largest time marker t min+1 the next lower brightness H max-1 , are assigned, and so on. This assignment of relative brightness values ​​H max , H max-1 , H max-2 , ...preferably continues until all pixels 2, in particular all pixels 2 for the 33 timestamps stored in memory, have been updated. i are stored, a brightness H i is assigned. The inventive method advantageously takes advantage of the fact that the brightness Hiv;ih inversely proportional to the time mark t i is. Optionally, the brightness can be adjusted. iv; ih also be standardized.

[0045] To determine a distribution of brightness H iv;ih The individual pixel 2 can be within a column S1, S2, ..., S x in a further process step, the respective brightness H iv;ih each with a timestamp t i provided pixel 2, preferably again by the computing unit 35, with a vertical position i v The respective pixel is multiplied by 2 and a sum Σ is calculated. iv i v * H iv;ih all products obtained in this way from brightness H iv;ih and vertical position i v be formed. The vertical position i v This can be done in particular in the form of the number of line Z1, Z2, ..., Z x The sum obtained Σ iv i v * H iv;ih can then be calculated by a sum Σ iv Hiv;ih all brightness levels H iv;ih the one with a timestamp t i pixel 2 of the respective column S1, S2, ..., S x to be divided, thereby achieving a subpixel-accurate vertical position COG v a center of brightness H iv;ih within the respective column S1, S2, ..., S x can be determined. Alternatively or cumulatively, Z1, Z2, ..., Z can be entered within a single line. x the respective brightness H iv;ih each with a timestamp t i pixel 2 with a horizontal position i h The respective pixel's value is multiplied by 2. A sum Σ ih i h * H iv;ih all products obtained in this way from brightness H iv;ih and horizontal position i h can then be formed and the resulting sum Σ ih i h * H iv;ih by the sum Σ ih H iv;ih all brightness levels H iv;ihwithin the respective line Z1, Z2, ..., Z x can be divided. This allows for advantageous subpixel-accurate horizontal positioning (COG). h a center of brightness H iv;ih within the respective line Z1, Z2, ..., Z x can be determined. Particularly in flying spot applications, the position of the center of luminance H can be advantageously determined in this way. iv;ih in both dimensions, i.e., both the vertical position COG v as well as the horizontal position COG h , can be determined with subpixel precision. Furthermore, a calculation in 3D coordinates can also be performed based on each of these variants.

[0046] Fig.Figure 3 shows, by way of example, the path of a laser line L across a section of the image area of ​​a CMOS image sensor 1 according to the invention, as well as, in an enlarged view, a section of the first column S1 and the associated intensity profile Int of the laser line L in said section. In the example shown here, the laser line L within the first column S1 covers only pixels 2 in rows Z7 to Z8. 12 from, where the intensity curve Int is Gaussian with a maximum in the area of ​​pixel 2 in rows Z9 and Z 10 Thus, the photon incidence on pixel 2 in rows Z9 and Z is also 10 The largest column is S1 in the first column and increases with both smaller (< Z8) and larger (> Z) values. 11 ) row numbers. In the example shown, pixels 2 would thus be at positions i = (S1; Z9) and (S1; Z9). 10 ) first reach a predetermined threshold value T and, according to the invention, the smallest time markers ti Pixel 2 is assigned. Depending on when the threshold value T is reached, the other pixels would follow. Pixel 2 in rows Z1 to Z6 and > Z 13 In the example shown, these would be unexposed and would advantageously remain disregarded within the framework of the inventive method in order to optimize the efficiency of the readout and data processing process.

[0047] Finally, to define the path of a laser line L across the regular arrangement of pixels 2 in columns S1, S2, ..., S x and lines Z1, Z2, ..., Z x To determine the vertical position COG of the CMOS image sensor 1 v of the center of brightness H iv;ih for multiple columns S1, S2, ..., S x , preferably for each individual column S1, S2, ..., S x , as previously described, can be determined. A connection of the various vertical positions COG vThis advantageously provides the path of the laser line L in the image area of ​​the CMOS image sensor 1.

[0048] The present invention relates to a method for the fast and energy-efficient operation of a CMOS image sensor 1, which can be used in particular in laser triangulation or similar methods that project a light line with a Gaussian intensity profile onto the CMOS image sensor 1. The method according to the invention is characterized in particular by the fact that a brightness H iv;ih The individual pixels 2 are based solely on timestamps stored in a memory 33. i This is determined so that overexposure of pixel 2, and thus of the CMOS image sensor 1, is avoided, and its dynamic range is increased accordingly. Furthermore, only pixel information (time stamp t) is recorded. i and position i v ; i hThe image sensor 1 processes the exposure of pixels 2 whose exposure has reached a predetermined threshold T, while unexposed pixels 2 are disregarded. Since the power consumption of an image sensor is proportional to the data throughput, the data processing and transmission optimized according to the invention not only lead to an increase in the scan rate, but also to lower power consumption and thus to particularly energy-efficient operation of the CMOS image sensor 1. Reference symbol list 1 CMOS image sensor 11. Row access circuit 12-column access circuit 2 pixels 21 Photodetector 22 Photocurrent 23 photoelectric switches 24 Integration capacitor 25 Comparator 3 Control unit 31 Counting unit (global counter) 32 clock signal 33 storage 34 Data readout unit 35 computing unit COG vPosition of the brightness center within a column (S1, S2, ..., S x ) COG h Position of the brightness center within a row (Z1; Z2; ...; Z x ) H iv;ih Brightness at position (i v ; i h) H max maximum brightness H max-1 , H max-2 H max-n Brightness values ​​(in order of decreasing brightness) i v vertical position of the pixel (2) i h horizontal position of the pixel (2) Σ iv i v * H iv;ih Sum of products from vertical position (i v ) and brightness (H iv; ih ) Σ ih i h * H iv; ih Sum of products from horizontal position (i h ) and brightness (H iv; ih ) L laser line R Reset circuit Z1, Z2, ..., Z xRow of pixels (2) of the CMOS image sensor (1) S1, S2, ..., S x Column of pixels (2) of the CMOS image sensor (1) T threshold value t i Time stamp t min smallest time mark (corresponds to the shortest time until the threshold value (T) is reached) t min+1 , t min+2 , ..., t min+n Time markers (each representing a longer time until the threshold (T) is reached)

Claims

[1] Method for operating a CMOS image sensor (1), wherein the CMOS image sensor (1) comprises at least: - a regular arrangement of pixels (2) in columns (S1, S2, ..., S x ) and rows (Z1, Z2, ..., Z x ); - wherein the pixels (2) each comprise at least one photodetector (21) and at least one comparator (25); - and a control unit (3), wherein the control unit (3) - a counting unit (31) for processing a clock signal (32) and for generating time markers (t i ), - and includes at least one storage (33); - and wherein the control unit (3) is configured to specify at least one threshold value (T) for a photon incidence; comprising at least the following procedural steps: - Resetting a pixel voltage of the photodetectors (21) of all pixels (2) to a defined initial value; - Reset all timestamps (t i) to zero; - Clearing the memory (33); - Starting the clock signal (32) at the start of an exposure of the CMOS image sensor (1); - Detecting each photon incidence using the photodetector (21) of the respective pixel (2); - Comparing an actual value of the detected photon incidence of each pixel (2) with the threshold value (T) for the photon incidence specified by the control unit (3) using the respective comparator (25) of the respective pixel (2); - if the threshold (T) is exceeded at a pixel (2) generate a timestamp (t) i ) for this pixel (2) by the counting unit (31); - Storing the timestamp (t i ) for this pixel (2) in the memory (33); - Reading the memory (33); and - Determining a brightness (H iv;ih ) of the individual pixels (2) based on the timestamps stored in memory (33) (t i ). [2] Method according to claim 1, wherein the storage device (33) - at least one column storage, in particular a column storage operated according to the FIFO principle, and in which the time stamp (t) generated for a specific pixel (2) i ) together with a vertical position (i v ) of pixel (2), in particular with the respective number of the row (Z1, Z2, ..., Z x ) of the CMOS image sensor (1) of the respective pixel (2) in which column memory is stored; and / or - at least one row memory, in particular a row memory operated according to the FIFO principle, and in which the time stamp (t) generated for a specific pixel (2) is stored i ) together with a horizontal position (i h ) of pixel (2), in particular with the respective number of the column (S1, S2, ..., S x ) of the CMOS image sensor (1) of the respective pixel (2), in which line memory is stored. [3] Method according to claim 1, wherein the memory (33) is at least a global memory, in particular a global memory operated according to the FIFO principle, and wherein the time stamp (t) generated for a particular pixel (2) i ) together with the horizontal (i h ) and the vertical (i v ) Position of the respective pixel (2) on the CMOS image sensor (1), in particular together with the respective number of the column (S1, S2, ..., S x ) and the respective line number (Z1, Z2, ..., Z x ) of the CMOS image sensor (1) of the respective pixel (2), in which global memory is stored. [4] Method according to claim 3, wherein, prior to reading the memory (33), the stored value triples comprising a time stamp (t) i ) and associated horizontal (i h ) and vertical (i v ) Position of the respective pixel (2) according to columns (S1, S2, ..., S x ) or rows (R1, R2, ..., R x) are sorted. [5] Method according to one or more of the preceding claims, wherein the control unit (3) specifies two or more threshold values ​​(T) and each pixel (2) comprises two or more comparators (25) corresponding to these, in particular one comparator (25) for each threshold value (T) specified by the control unit (3). [6] A method according to one or more of the preceding claims, wherein the generation and storage of time stamps (t i ) for pixels (2) within a column (S1, S2, ..., S x ) is terminated after a predetermined number of pixels (2), in particular after 2, 3, 4, 5, 6, or more pixels (2), wherein the predetermined number of pixels (2) is preferably even, in particular 2, 4 or 6 pixels (2). [7] Method according to one or more of the preceding claims, wherein the generation and storage of timestamps (t i) for pixels (2) from the control unit (3) only for a previously defined range within a column (S1, S2, ..., S x ), especially only for previously defined rows (Z1, Z2, ..., Z x ) within a column (S1, S2, ..., S x ), is admitted. [8] Method according to one or more of the preceding claims, wherein to determine the brightness (H iv;ih ) of the individual pixels (2) of a column (S1, S2, ..., S x ) - first the pixel (2) with the smallest timestamp (t min ) determined and designated as pixel (2) with the greatest brightness (H max ) is classified; - then the one corresponding to pixel (2) with the greatest brightness (H max ) adjacent pixels (2) within a column (S1, S2, ..., S x ) are each compared with each other and the pixel (2) with the smallest time stamp (t) is assigned to the pixel (2). min ) next largest time marker (t min+1) the next lower brightness (H max-1 ), is assigned and so on. - to all pixels (2), in particular all pixels (2) for the timestamps (t) in memory (33) i ) are stored, a brightness (H iv;ih ) is assigned; and / or to determine brightness (H iv;ih ) of the individual pixels (2) of a row (Z1, Z2, ..., Z x ) - first the pixel (2) with the smallest timestamp (t min ) determined and designated as pixel (2) with the greatest brightness (H max ) is classified; - then the one corresponding to pixel (2) with the greatest brightness (H max ) adjacent pixels (2) within a row (Z1, Z2, ..., Z x ) are each compared with each other and the pixel (2) with the smallest time stamp (t) is assigned to the pixel (2). min ) next largest time marker (t min+1 ) the next lower brightness (H max-1 ), is assigned and so on. - to all pixels (2), in particular all pixels (2) for the timestamps (t) in memory (33) i ) are stored, a brightness (H iv;ih ) is assigned. [9] Method according to claim 8, wherein to determine a distribution of brightness (H iv;ih ) of the individual pixels (2); - within a column (S1, S2, ..., S x ): - the respective brightness (H iv;ih ) each with a timestamp (t i ) provided pixel (2) with a vertical position (i v ) of the respective pixel (2) is multiplied; - a sum (Σ iv i v * H iv;ih ) all products obtained so from brightness (H iv;ih ) and vertical position (i v ) is formed; - the sum received (Σ iv i v * H iv;ih ) by a sum (Σ iv H iv;ih ) of all brightness levels (H iv;ih ) the one with a time stamp (t i) provided pixel (2) of the respective column (S1, S2, ..., S x ) is divided; - and thus a subpixel-accurate vertical position (COG) v ) of a center of brightness (H iv;ih ) within the respective column (S1, S2, ..., S x ) is determined; and / or - within a row (Z1, Z2, ..., Z x ): - the respective brightness (H iv;ih ) each with a timestamp (t i ) provided pixels (2) with a horizontal position (i h ) of the respective pixel (2) is multiplied; - a sum (Σ ih i h * H iv;ih ) all products obtained so from brightness (H iv;ih ) and horizontal position (i h ) is formed; - the sum received (Σ ih i h * H iv;ih ) by the sum (Σ ih H iv;ih ) of all brightness levels (H iv;ih ) within the respective row (R1, R2, ..., Rx ) is divided; - and thus a subpixel-accurate horizontal position (COG) h ) of a center of brightness (H iv;ih ) within the respective row (R1, R2, ..., R x ) is determined. [10] Method according to claim 9, wherein a laser line (L) is traced across the regular arrangement of pixels (2) in columns (S1, S2, ..., S x ) and rows (Z1, Z2, ..., Z x ) of the CMOS image sensor (1) is determined by the vertical position (COG) v ) of the center of amplitude of brightness (H iv;ih ) for multiple columns (S1, S2, ..., S x ), preferably for each individual column (S1, S2, ..., S x ). is determined. [11] Method according to one or more of the preceding claims, wherein the exposure of the respective pixel (2) is interrupted when the threshold value (T) is reached, in particular by opening a photocurrent switch (23) associated with the respective pixel (2), wherein preferably an integrating capacitor (24) associated with the respective pixel (2) is discharged by means of a reset circuit (R) associated with the respective pixel (2). [12] CMOS image sensor (1), at least comprising: - a regular arrangement of pixels (2) in columns (S1, S2, ..., S x ) and rows (Z1, Z2, ..., Z x ); - wherein the pixels (2) each comprise at least one photodetector (21) and at least one comparator (25); - and a control unit (3), - wherein the control unit (3) includes a counting unit (31) for processing a clock signal (32) and for generating time markers (t i ), - and includes at least one storage (33); - and wherein the control unit (3) is configured to specify at least one threshold value (T) for a photon incidence, characterized by , that the control unit (3) is configured to execute the method according to any one of claims 1 to 11. [13] CMOS image sensor (1) according to claim 12, characterized by , that each pixel (2) includes two or more comparators (25). [14] CMOS image sensor (1) according to claim 12 or 13, characterized by , that the CMOS image sensor (1) is designed as a BSI sensor. [15] Use of a CMOS image sensor (1) according to one of claims 12 to 14 in object measurement and / or object inspection methods in which only a part of the pixels (2) of a CMOS image sensor (1) is exposed, in particular - in methods in which a light line with a Gaussian intensity profile is imaged onto the CMOS image sensor (1); or - in methods in which a single light point or several single light points are imaged onto the CMOS image sensor (1).