DE102021212659A1 - Process for processing measured values of an optical sensor arrangement - Google Patents

Abstract

The invention relates to a method for processing measured values (M1, M2) of an optical sensor arrangement (1), wherein at least two measured values (M1, M2), which are recorded by the optical sensor arrangement (1) as raw data, at least one property (m1 ,2) is extracted, which is output as the result of the processing for further evaluation, with a ratio of the measured values (M1, M2) being formed to extract the at least one property (m1, 2), with the arrangement of the two measured values (M1, M2) is carried out when forming the ratio depending on a comparison of the two measured values (M1, M2), with the smaller measured value of the two measured values (M1, M2) being set in the counter when the ratio is formed, and the larger measured value of the two measured values ( M1, M2) is placed in the denominator when forming the ratio, with a substitute value being output as the at least one property (m1,2) if both measured values (M1, M2) correspond to the value "zero", as well as an optical sensor arrangement (1st ) and a computer program product for carrying out the method.

Description

The invention relates to a method for processing measured values from an optical sensor arrangement. The present invention also relates to an optical sensor arrangement and a computer program product for carrying out such a method for processing measured values.

In order to ensure safe, highly automated driving, the coefficient of friction of the road to be traveled should be known as precisely as possible. The coefficient of friction is particularly influenced by the "intermediate media" (media between asphalt and tires) that occur on the road, such as water, snow, ice, leaves, oil, etc. These media can be detected by various sensors, such as cameras, sound sensors, ultrasonic sensors or by optical sensors in the infrared range.

Optical sensors in the near infrared spectrum (approx. 800nm - 3000nm) evaluate a diffuse and/or directed reflection of actively emitted light of several wavelength ranges. These wavelength ranges differ in that they contain differently pronounced absorption lines of water in all states of aggregation (liquid, ice-covered, snow-covered or mixed states). This means that dry roads can be distinguished from non-dry ones, individual intermediate media can be categorized as “road conditions” and even the layer thicknesses of the intermediate medium can be determined. For this purpose, intensities and intensity ratios in particular are evaluated, either via comparisons, threshold values or machine learning methods.

From the DE 27 12 199 B2 a device for warning the driver of a motor vehicle of slippery roads is known. The device comprises a light source, through which light of a specific wavelength can be emitted onto the road surface, a receiver, through which the light reflected from the road can be converted into a corresponding electrical signal, and a warning device, through which the ice forms on the road resulting electrical signal the motor vehicle driver is recognizable.

From the DE 41 33 359 C2 a method for measuring the thickness of a layer of water present on a roadway is known, in which a light beam with a limited spatial angle and an infrared spectral component is emitted onto the roadway by means of a light source. Light scattered back from the roadway, which has penetrated the existing water layer, is selectively scanned in predetermined solid angle ranges for at least two wavelengths lying in the near infrared range. The thickness of the water layer is determined from the amplitudes of the selectively sampled, backscattered luminous fluxes and displayed by means of a display unit.

From the EN19506550 is a known method for the instantaneous detection and warning of dangers on traffic routes caused by ice formation, such as, for example, freezing wetness. Ice forming on aircraft and machine components that are not protected against frost should also be detected in good time. The degree of crystallization of the molecular structure present in the liquid and/or solid aggregate state is determined by spectral analysis and computer-aided chemometric evaluation of the transmission and reflection properties of water or a protic liquid or solution that behaves like water and is capable of solidifying. The spectral analysis measurements take place in one or more wavelength ranges between 700nm and 10µm, whereby discrete wavelength ranges can also be used for the evaluation of the spectral properties.

Disclosure of Invention

The method for processing measured values of an optical sensor arrangement with the features of independent claim 1 has the advantage that by extracting at least one property from the at least two measured values as raw data, the measured values are processed and transformed by an upstream algorithm or mathematical operations in such a way that that subsequent machine learning algorithms as well as classic methods, such as a comparison with threshold values, can be improved or simplified.

Embodiments of the method according to the invention for processing measured values of an optical sensor arrangement are particularly suitable for an optical sensor arrangement in a vehicle for road condition detection. However, the method can also be used with all other sensor arrangements with optical sensors, which can be used with a number of measured values, for example with different NEN wavelengths or directions of polarization work. Areas of application are, for example, a multi-wavelength lidar, a mini spectrometer for mobile phones or industrial line camera systems for monitoring, controlling or regulating production processes.

In the case of optical sensor arrangements known from the prior art, the at least one optical sensor measures intensities of light at different wavelengths in the infrared spectrum in order to determine the state of the road. The resulting measured values are further processed by algorithms, often machine learning algorithms such as SVM or neural networks, in order to make a statement about the road condition. By means of embodiments of the method according to the invention for processing measured values of an optical sensor arrangement, these measured values can be processed as raw data using mathematical operations and/or transformations in such a way that the susceptibility to interference and any falsification of the determined road condition that may occur as a result can be at least reduced or, ideally, completely eliminated .

Due to the processing of the measured values according to the invention, the downstream algorithms require less complexity than without processing or after processing known from the prior art. This leads to a more cost-effective implementation for the product by saving computing power and/or storage capacity. In addition, it is possible to increase the quality of the output of the optical sensor arrangement, so that a better hit rate when classifying surfaces and/or greater accuracy when calculating the layer thickness can be achieved. Embodiments of the method according to the invention for processing measured values of an optical sensor arrangement are not very computationally intensive and can be implemented without special hardware. In addition, the applicability of the method according to the invention for processing measured values from an optical sensor arrangement is not limited to road condition detection, but can in principle be applied to all types of optical sensor arrangements in which at least one relative value of at least two measured values is processed.

Embodiments of the present invention provide a method for processing measured values of an optical sensor arrangement. According to the method according to the invention, at least one property is extracted from at least two measured values, which are recorded by the optical sensor arrangement as raw data, and which is output as the result of the processing for further evaluation. In this case, a ratio of the measured values is formed to extract the at least one property, with the arrangement of the two measured values being carried out when forming the ratio as a function of a comparison of the two measured values, with the smaller measured value of the two measured values being set in the counter when forming the ratio, and the larger measured value of the two measured values is placed in the denominator when the ratio is formed, with a substitute value being output as the at least one property if both measured values correspond to the value “zero” or “0”.

In addition, an optical sensor arrangement with at least one optical sensor and an evaluation and control unit is proposed, which is set up to carry out the method according to the invention for processing measured values.

By outputting the substitute value, the mathematically undefined case can be intercepted in which both measured values correspond to the value "zero" or "0". Embodiments of the present invention can be combined with other methods in parallel or in series, for example to generate multiple properties for subsequent algorithms. A recursive application of the processing to more than two measured values is also possible.

In the present case, the evaluation and control unit can be understood to mean an electrical device, such as a control unit, which processes or evaluates detected sensor signals. The evaluation and control unit can have at least one interface, which can be in the form of hardware and/or software. In the case of a hardware design, the interfaces can be part of a so-called system ASIC, for example, which contains a wide variety of functions of the evaluation and control unit. However, it is also possible for the interfaces to be separate integrated circuits or to consist at least partially of discrete components. In the case of a software design, the interfaces can be software modules which are present, for example, on a microcontroller alongside other software modules. A computer program product with program code is also advantageous, which is stored on a machine-readable medium such as a semiconductor memory, a hard disk memory or an optical memory and is used to process the measured value when the program is executed by the evaluation and control unit.

The measures and developments listed in the dependent claims allow advantageous improvements to the method specified in independent patent claim 1 for processing measured values of an optical sensor arrangement.

In an advantageous embodiment of the method according to the invention, the substitute value can be selected from the values (-1, 0, 1), for example. Of course, other substitute values can also be used in order to clarify the case for subsequent components of the measured value processing.

In a further advantageous embodiment of the method according to the invention, at least the calculated ratio of the two measured values can be output as the at least one property. By arranging the smaller measured value in the numerator, the value range of the ratio can be limited to the enclosed interval between 0 and 1 without significant loss of information.

In a preferred embodiment of the method according to the invention, the calculated ratio can be subtracted from the value “1” and at least the subtraction result can be output as the at least one property. As a result, if there are small differences between the two measured values, jumps in the output property can advantageously be avoided and the continuity of the at least one output property can be maintained.

In a further advantageous embodiment of the method according to the invention, the recorded first measured value and/or the recorded second measured value can each be set to a specified value if they fall below a specified threshold value. Such a "noise threshold" can improve the noise behavior of the evaluation method.

In a further advantageous embodiment of the method according to the invention, the recorded measured values and/or the calculated ratio of the two measured values can be subjected to a logarithmic transformation. As a result, the processing of the measured values can be adapted to the corresponding subsequent evaluation processes.

In a further advantageous embodiment of the method according to the invention, the two measured values can be compared by forming a difference, in which a second measured value is subtracted from a first measured value. In this case, the sign of the difference formed can determine the arrangement of the measured values when the ratio is formed, it being possible for the sign of the difference formed to be output as part of the at least one property. If the difference formed has a negative sign, the first measured value can be placed in the numerator and the second measured value in the denominator.

If the difference formed has a positive sign, the second measured value can be placed in the numerator and the first measured value in the denominator. As a result, the value range of the at least one extracted property can be limited to the enclosed interval between -1 and 1 without significant loss of information, which is relevant for many machine learning algorithms. The information as to whether the first measured value or the second measured value is greater can be read simply from the sign of the difference.

Embodiments of the invention are shown in the drawings and are explained in more detail in the following description. In the drawings, the same reference symbols denote components or elements that perform the same or analogous functions.

character list

• 1 shows a schematic flowchart of an exemplary embodiment of a method according to the invention for processing measured values of an optical sensor arrangement.
• 2 shows a schematic block diagram of an exemplary embodiment of an optical sensor arrangement according to the invention for carrying out the method according to the invention 1 .

Embodiments of the invention

How out 1 and 2 As can be seen, the illustrated exemplary embodiment of a method 100 according to the invention for processing measured values M1, M2 of an optical sensor arrangement 1 comprises a step S100 in which at least two measured values M1, M2 from the optical sensor arrangement 1 recorded as raw data. In a step S110, the two measured values M1, M2 are compared with one another. In a step S120, a ratio V of the two measured values M1, M2 is formed in order to extract at least one property m1,2. For this purpose, the smaller measured value (min(M1, M2)) of the two measured values M1, M2 is placed in the numerator to form the ratio, and the larger measured value (max(M1, M2)) of the two measured values M1, M2 is used in the denominator to form the ratio set. In step S130, the extracted at least one property m1,2 is output as the result of the processing for further evaluation, with a substitute value a being output as the at least one property m1,2 if both measured values M1, M2 have the value “zero” or “zero”. 0".

The method 100 for processing measured values M1, M2 of an optical sensor arrangement 1 can be implemented, for example, in software or hardware or in a mixed form of software and hardware in an evaluation and control unit 10.

How out 2 It can also be seen that the illustrated exemplary embodiment of the optical sensor arrangement 1 according to the invention comprises at least one optical sensor 12 for acquiring at least two measured values M1, M2 and an evaluation and control unit 10 which is set up to carry out the method 100 according to the invention.

In order to minimize the influence of interference variables on the measured signal, it is usual with optical sensors 12 to form a relative value or a ratio V from two measured values M1, M2 by dividing them by one another. In the exemplary embodiment shown, a reflectance or degree of reflection at two different wavelengths, for example, is recorded as measured values M1, M2, which describes the relationship between reflected and incident intensity of the electromagnetic waves or light waves. The situation can arise in this case that at least one of the two measured values M1, M2 assumes the value “zero” or “0” or becomes numerically small close to “zero” or “0”. This is the case, for example, with strong absorption. If, for example, the measured value M2 in the denominator approaches “zero” or “0” with a constant measured value M1 in the numerator, the ratio V tends to infinity. If the measured value M2 assumes the value "zero" (M2 = 0) in the denominator, then the ratio V is not mathematically defined. This behavior is problematic for subsequent algorithms since the ratio V of the two measured values M1, M2 is not within fixed limits or is even undefined. If it is known that only one of the two measured values M1, M2 is small, the arrangement of the measured values M1, M2 for the ratio formation or division can be selected in such a way that the smaller measured value is in the numerator, so that the ratio V at decreasing measured value M1, M2 does not tend towards infinity. However, if this is not possible because, for example, both measured values M1, M2 can be numerically small or "zero" or "0", or because it is not known which measured value M1, M2 approaches "zero" or "0", then this is the case this can be solved by embodiments of the method 100 according to the invention for processing measured values M1, M2 of an optical sensor arrangement 1.

Equation (1) shows a first exemplary embodiment for processing the measured values M1, M2 according to the invention and for extracting the at least one property m1,2 from the two measured values M1, M2. $$\text{m}1.2=\text{V}=\text{at least}\left(\text{M}\mathrm{1,}\text{M}2\right)/\text{Max}\left(\text{M}\mathrm{1,}\text{M}2\right)$$

Here, the "min" operator returns the minimum of the arguments and the "max" operator returns the maximum of the arguments. If both measured values M1, M2 have the same value not equal to "zero" or "0", then the at least one extracted property m1,2 corresponds to the value "one" or "1". If both measured values M1, M2 correspond to the value "zero" or "0", then the substitute value a is output as the at least one property m1,2. For example, the value "minus one" or "-1" or the value "zero" or "0" can be output as the substitute value a.

In order to prevent the at least one property m1,2 from jumping, a second exemplary embodiment shown in equation (2) is preferably used for processing the measured values M1, M2 according to the invention and for extracting the at least one property m1,2 from the two measured values M1, M2 used. $$\text{m}1.2=(1-\left(\text{V}\right)=(1-(\text{at least}\left(\text{M}\mathrm{1,}\text{M}2\right)/\text{Max}\left(\text{M}\mathrm{1,}\text{M}2\right))$$

Here, the "min" operator returns the minimum of the arguments and the "max" operator returns the maximum of the arguments. If both measured values M1, M2 have the same value not equal to "zero" or "0", then the at least one extracted property m1,2 corresponds to the value “zero” or “0”. If both measured values M1, M2 correspond to the value "zero" or "0", then the substitute value a is output as the at least one property m1,2. For example, the value "minus one" or "-1" or the value "one" or "1" can be output as the substitute value a.

$$\text{sgn}\left(\text{D}\right)=\text{sgn}\left(\text{M}1-\text{M}2\right)$$

In the exemplary embodiment shown, the two measured values M1, M2 are compared by forming a difference D according to equation (3), in which a second measured value M2 is subtracted from a first measured value M1. $$\text{D}=\left(\text{M}\mathrm{1,}\text{M}2\right)$$

$$\text{so called}\left(\text{D}\right)=\text{so called}\left(\text{M}1-\text{M}2\right)$$

The sign of the formed difference D, which the operator "sgn" supplies from Equation (4), determines the arrangement of the measured values M1, M2 when forming the ratio V. Combining Equation (4) with Equation (1) results in a third exemplary embodiment shown in equation (5) for processing the measured values M1, M2 according to the invention and for extracting the at least one property m1,2 from the two measured values M1, M2. $$\text{m}1.2=(\text{so called}\left(\text{D}\right))\left(\text{V}\right))=(\text{so called}\left(\text{M}1-\text{M}2\right)(\text{at least}\left(\text{M}\mathrm{1,}\text{M}2\right)/\text{Max}\left(\text{M}\mathrm{1,}\text{M}2\right))$$

In this case, the sign of the difference D formed is output in addition to the ratio V as part of the at least one property m1,2, as can be seen from equation (5). If the difference D formed has a negative sign, the first measured value M1 is set in the numerator and the second measured value M2 is set in the denominator. If the difference D formed has a positive sign, the second measured value M2 is set in the numerator and the first measured value M1 is set in the denominator.

The combination of equation (4) with equation (2) results in a fourth exemplary embodiment shown in equation (6) for processing the measured values M1, M2 according to the invention and for the extraction of the at least one property m1,2 from the two measured values M1, M2. $$\text{m}1.2=(\text{so called}\left(\text{D}\right))\left(1-\text{V}\right))=(\text{so called}\left(\text{M}1-\text{M}2\right)(1-(\text{at least}\left(\text{M}\mathrm{1,}\text{M}2\right)/\text{Max}\left(\text{M}\mathrm{1,}\text{M}2\right)))$$

In this case, the sign of the difference D formed is also output as part of the at least one property m1,2, as can be seen from equation (6). If the difference D formed has a negative sign, the first measured value M1 is set in the numerator and the second measured value M2 is set in the denominator. If the difference D formed has a positive sign, the second measured value M2 is set in the numerator and the first measured value M1 is set in the denominator.

The "sgn" operator returns the sign of the argument, the "min" operator returns the minimum of the arguments, and the "max" operator returns the maximum of the arguments. The arrangement of the ratio formation is thus carried out depending on the comparison of the two measured values M1, M2. The arrangement information is displayed via the sign of the result. This limits the value range of the at least one property m1,2 to the enclosed interval between -1 and 1 without significant loss of information, which is relevant for many machine learning algorithms. The information as to whether the first measured value M1 or the second measured value M2 is greater can be read off simply from the sign of the at least one property m1,2. If this is not relevant for the application, the operator "sgn" can also be omitted as an embodiment.

Equations (1) to (6) can be applied to all possible combinations of the individually measured measured values M1, M2 in order to form the at least one property m1,2. The embodiment with reversed order of subtraction in the operator "sgn" only changes the sign of the result of the processing and is to be regarded as an equivalent embodiment.

To define a noise threshold, the recorded first measured value M1 and/or the recorded second measured value M2 can each be set to a specified value if they fall below a specified threshold value. In addition, the recorded measured values M1, M2 and/or the calculated ratio V of the two measured values M1, M2 can optionally be subjected to a logarithmic transformation.

The processing of measured values M1, m2 described can also be implemented in several program branches instead of a mathematical equation via a case distinction based on the measured values M1, M2.

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• DE 2712199 B2 [0004]
• DE 4133359 C2 [0005]
• DE 19506550 [0006]

Claims (11)

Method (100) for processing measured values (M1, M2) of an optical sensor arrangement (1), at least one property (m1, 2) is extracted, which is output as the result of the processing for further evaluation, wherein a ratio (V) of the two measured values (M1, M2) is formed to extract the at least one property (m1, 2), the arrangement of the two measured values (M1, M2) is carried out when forming the ratio depending on a comparison of the two measured values (M1, M2), with the smaller measured value (min(M1, M2)) of the two measured values (M1, M2) being included in the numerator when forming the ratio is set, and the larger measured value (max(M1, M2)) of the two measured values (M1, M2) is set in the denominator when forming the ratio, with a substitute value (a) being output as the at least one property (m1,2). , if both measured values (M1, M2) correspond to the value "zero". Method (100) according to claim 1 , characterized in that the replacement value (a) is selected from the values (-1, 0, 1). Method (100) according to claim 1 or 2 , characterized in that at least the calculated ratio (V) of the two measured values (M1, M2) is output as the at least one property (m1,2). Method (100) according to claim 1 or 2 , characterized in that the calculated ratio (V) of the two measured values (M1, M2) is subtracted from the value “1” and at least the subtraction result is output as the at least one property (m1,2). Method (100) according to any one of Claims 1 until 4 , characterized in that the detected first measured value (M1) and/or the detected second measured value (M2) are each set to a predetermined value if they fall below a predetermined threshold value. Method (100) according to any one of Claims 1 until 5 , characterized in that the measured values (M1, M2) and/or the calculated ratio (V) of the two measured values (M1, M2) are subjected to a logarithmic transformation. Method (100) according to any one of Claims 1 until 6 , characterized in that the two measured values (M1, M2) are compared by forming a difference (D), in which a second measured value (M2) is subtracted from a first measured value (M1). Method (100) according to claim 7 , characterized in that the sign of the formed difference (D) determines the arrangement of the measured values (M1, M2) when forming the ratio (V), the sign of the formed difference being output as part of the at least one property (m1,2). becomes. Method (100) according to claim 8 , characterized in that with a negative sign of the formed difference (D), the first measured value (M1) is placed in the numerator and the second measured value (M2) in the denominator, and with a positive sign of the formed difference (D) of second measured value (M2) is placed in the numerator and the first measured value (M1) in the denominator. Optical sensor arrangement (1) with at least one optical sensor (12) and an evaluation and control unit (10), which is set up, the method according to one of Claims 1 until 9 to execute. Computer program which is set up, the method according to one of Claims 1 until 9 to execute.

Images