Detection method for detecting alarm objects in pieces of luggage

Abstract

The present invention relates to a detection method for detecting alarm objects (AO) in pieces of luggage (G), including the following steps: - performing a transmission analysis (TA) on a piece of luggage (G) with creation of a transmission result (TE), - segmenting the transmission result (TE) for the purpose of identifying objects (O), - creating a simulated simulation detector signal (SDS) for at least one identified object (O) with assignment of an alarm material (AM) by means of a diffraction simulation (DS) specific to a diffraction analysis (DA), - performing a diffraction analysis (DA) on the same piece of luggage (G) with creation of a measured detector signal (MDS), - reconstructing a measured diffraction result (MDE) from the created measured detector signal (MDS) for the at least one object (O),- reconstructing a simulation diffraction result (SDE) with the same dimensionality as the measured diffraction result (MDE) on the basis of the created simulation detector signal (SDS) for the at least one object (O), - comparing the measured diffraction result (MDE) with the simulation diffraction result (SDE), - outputting a comparison result (VE).

Description

[0001] Detection method for detecting alarm objects in luggage

[0002] The present invention relates to a detection method for detecting alarm objects in luggage and to a computer program product comprising instructions for executing such a detection method.

[0003] It is known that baggage screening should be carried out in security-sensitive areas, such as airports. At airports, this applies in particular to so-called checked baggage, i.e., pieces of baggage that are to be transported in the cargo hold of an aircraft. In known solutions, the baggage is usually subjected to a transmission analysis. One such transmission analysis is an X-ray computed tomography procedure. The result of such a transmission procedure can be referred to as a three-dimensional transmission result and serves to differentiate between different objects within the baggage and to at least partially determine their degree of danger. In known security systems, a decision is made based on the transmission analysis and the evaluation of the three-dimensional transmission result as to whether or not the baggage may contain an alarm object.This can also be understood as the output of an alarm signal.

[0004] To ensure that not every piece of luggage in which a possible alarm object has been found actually has to be opened and checked manually, many airports use a second, downstream detection method. This usually involves a visual inspection of the transmission result by a person or possibly an algorithm. In some cases, a diffraction analysis is also carried out using an X-ray diffraction device. Every piece of luggage that has triggered an alarm is subjected to such a diffraction analysis. Here, the luggage is analyzed using X-rays. X-rays diffracted in the luggage are recorded by a corresponding detector and output as a detector signal. The detector signal itself is only partially meaningful and is used to reconstruct a diffraction result based on this detector signal.This reconstruction is typically performed by converting the detector signal from the detector coordinate system to the luggage coordinate system. In a final step, known methods compare the reconstructed diffraction result with information from an alarm database to determine whether the material of the alarm object is alarm material. In other words, the alarm can be either confirmed or resolved with regard to the material composition of the alarm object.

[0005] A disadvantage of the known solutions is that the alarm databases present the diffraction results, particularly in the form of momentum transfer functions, in an idealized manner. During the actual implementation of diffraction analysis, the detector signal is affected by a multitude of environmental parameters, luggage parameters, and device parameters of the diffraction device, so that significant deviations can sometimes occur between an idealized diffraction result from the alarm database on the one hand and the actual measurement result as a diffraction result on the other. For example, this deviation can arise from the absorption of parts of the radiation, so that the data contains photon noise. Another example is the mixing of signals from different materials.As a result, these differences lead to an increased number of false alarms, i.e. an unnecessary manual check of the baggage.

[0006] The object of the present invention is to at least partially remedy the disadvantages described above. In particular, the object of the present invention is to improve the measurement result in a cost-effective and simple manner and to avoid or at least reduce false alarms with a high degree of certainty.

[0007] The above object is achieved by a recognition method having the features of claim 1 and a computer program product having the features of claim 12. Further features and details of the invention emerge from the subclaims, the description, and the drawings. Features and details described in connection with the recognition method according to the invention naturally also apply in connection with the computer program product according to the invention, and vice versa, so that with regard to the disclosure of the individual aspects of the invention, reference is always made to each other.

[0008] According to the invention, a detection method is used to detect alarm objects in luggage. For this purpose, the detection method comprises the following steps:

[0009] Carrying out a transmission analysis on a piece of luggage to generate a transmission result,

[0010] Segmenting the transmission result to identify objects,

[0011] Generating a simulated simulation detector signal for at least one identified object with assignment of an alarm material by means of a diffraction simulation specific for a diffraction analysis,

[0012] Carrying out a diffraction analysis on the same piece of luggage to generate a measurement detector signal,

[0013] Reconstruction of a measurement diffraction result from the generated measurement detector signal for the at least one object,

[0014] Reconstruction of a simulation diffraction result with the same dimensionality as the measurement diffraction result on the basis of the generated simulation detector signal for the at least one object,

[0015] Comparison of the measurement diffraction result with the simulation diffraction result,

[0016] Output of a comparison result.

[0017] A recognition method according to the invention is based on known solutions in which a combination of two different recognition devices is used. As in the prior art, it is a combination of a transmission device for carrying out a transmission analysis and a diffraction device for carrying out a diffraction analysis. This combination is also referred to as a system of systems (SoS) device. As with known solutions, the recognition method starts with the performance of a transmission analysis, for example using a computer tomography device for checking all pieces of checked baggage at an airport. During this process, a transmission result is created for each piece of baggage; in many cases, this is an image of the density distribution.This transmission result, for example, a three-dimensional one, is then segmented, as is also already known, so that individual objects within the transmission result can be recognized as such and differentiated from other objects. This segmentation for object recognition can, for example, involve combining individual volume components, so-called voxels, which together form a homogeneous volume and can therefore be assigned with a high degree of probability to a single, specific object. As a result, one or more objects within this transmission result can be identified based on this transmission result.

[0018] The inventive concept is based on the fact that a simulation is now carried out on at least one identified object. This simulation is a diffraction simulation, i.e., a simulation for generating a simulation detector signal, as would be expected if precisely this object is subjected to a real diffraction analysis in a second step. In other words, the subsequent diffraction analysis is simulated in this step, and this simulation generates a simulation detector signal. This means that, for this object, at least one alarm material is assumed within the transmission result for the subsequent diffraction simulation. A projection into the future is thus made in the event that the object consists of, or at least contains, the assumed alarm material. This projection produces a simulated detector signal as could be expected if the diffraction analysis were actually carried out.The diffraction simulation thus simulates the actual analysis process of the subsequent diffraction analysis and thus makes a prediction based on at least one assumed alarm material.

[0019] Following this simulation step according to the invention, or even in parallel with it, the diffraction analysis can now actually be carried out on a diffraction device, generating a measured measurement detector signal. In the next two steps, as already explained in the introduction, a simulation diffraction result and a measurement diffraction result are generated from the respective detector signals, i.e., from the simulation detector signal and the measurement detector signal, via reconstruction steps. The reconstruction can, for example, involve the aforementioned backprojection from the detector coordinate system into the luggage coordinate system. The simulation diffraction result and the measurement diffraction result preferably have the same dimensionality, i.e., they are both three-dimensional.Other dimensionalities, for example carrying out the procedure with 2 dimensions or even more than 3 dimensions, are also conceivable in principle.

[0020] The core idea of ​​the invention is that in the final comparison step, the measurement diffraction result no longer needs to be compared with a standardized and, above all, idealized alarm database. Instead, a comparison is made with a specific and individual simulation diffraction result for the precisely identified object with an assumed alarm material. In other words, in the comparison step, not only the measurement diffraction result is individual for the current detection situation, but also the comparison value in the form of the simulation diffraction result. Finally, the comparison result is output, particularly in a qualitative manner, i.e., with a statement as to whether the object actually exhibits the assumed and assigned alarm material or not.

[0021] In contrast to conventional methods, an idealized alarm database is no longer used to make the final comparison as to whether an alarm object is present or not. Instead, the comparison value is also generated specifically and individually for each detected detection situation. This generation is ensured by the simulation step explained above, which ensures that specific characteristics and individual distortions in the diffraction analysis performed are taken into account during the diffraction simulation, even for assumed alarm materials. In other words, the comparison value is adapted to the individual situation of the object through the diffraction simulation, allowing it to better predict what an assumed alarm material for this object might actually look like in the form of a simulated detector signal.By preferably reconstructing the simulation diffraction result generated on this basis, detection accuracy is significantly increased, reducing the number of false alarms and significantly increasing the reliability of detecting alarm materials. This is also helped by the fact that the diffraction simulation can predict the noise levels that the diffraction measurement will exhibit. This allows statistical deviations to be distinguished from systematic deviations, which will increase detection performance.

[0022] It can be advantageous if, in a detection method according to the invention, a simulated simulation detector signal is additionally generated for adjacent regions next to the at least one identified object with an associated material. Since interaction with neighboring regions around the object also causes an impairment of the expected measurement detector signal, this impairment can already be taken into account in the diffraction simulation. Thus, not only the object but also neighboring volume sections, also referred to as neighboring voxels, are relevant for the measurement detector signal and must therefore also be taken into account in this embodiment for generating the simulation detector signal.In this way, it is possible to achieve improved accuracy in the simulation, so that the accuracy for the subsequent comparison and the associated reliability in the detection of alarm materials can be increased accordingly.

[0023] It may further be advantageous if, in a recognition method according to the invention, a preliminary analysis is carried out for at least one identified object, in particular with regard to at least one of the following parameters:

[0024] Shape of at least one identified object,

[0025] Density of at least one identified object.

[0026] The above list is not exhaustive. Of course, two or more parameters can also be considered in combination during such a preliminary analysis. These parameters are used to decide in the preliminary analysis whether the object could fundamentally be an alarm object or not. Therefore, such a preliminary analysis can also be referred to as a pre-alarm, which leads to the piece of luggage being forwarded to the second screening stage in the form of diffraction analysis. If the preliminary analysis excludes an alarm object with a high degree of probability, this piece of luggage can be classified as "safe" and bypassed by diffraction analysis. Furthermore, the preliminary analysis can provide an indication of what potential alarm material it might be.If, for example, shape, density or other parameters indicate that an explosive material is suitable as an alarm material, the alarm material for the generation step in the diffraction simulation can be selected accordingly based on this result of the preliminary analysis.

[0027] It can also be advantageous if, in a detection method according to the invention, the further analysis is carried out depending on the result of the preliminary analysis. In the simplest case, this preliminary alarm can therefore generally decide whether the piece of luggage is submitted for diffraction analysis or not. However, it is also possible to select a correspondingly adjusted alarm material or a defined quantity of alarm materials. The analysis settings of the diffraction device and / or the reconstruction mechanism can also be adjusted.

[0028] Furthermore, it may be advantageous if, in a detection method according to the invention, the diffraction simulation takes into account at least one of the following parameters:

[0029] Angular resolution of the diffraction analysis to be performed,

[0030] Spatial resolution of the diffraction analysis to be performed, which can lead to signal mixing,

[0031] Attenuation of the scattered signal of the diffraction analysis to be performed by other objects in the luggage and the associated expected distribution of the variance of the (Poisson) noise over the momentum transfer axis (energy axis)

[0032] Attenuation of the detector response during the diffraction analysis to be performed. The above list is non-exhaustive. The parameters to be considered depend particularly on and are selected for the diffraction device actually used. Such parameters can also be adjusted specifically for the respective piece of luggage and / or object in order to further increase the accuracy of the diffraction simulation result and thus improve the quality of a detection method according to the invention.

[0033] It may further be advantageous if, in a recognition method according to the invention, a backprojection or reconstruction is performed to generate the measurement diffraction result and / or the simulation diffraction result. This backprojection for the reconstruction step is, as already explained, for example, a backprojection from the coordinate system of the detector into the coordinate system of the piece of luggage. In particular, identical backprojection algorithms can be used for both the simulation diffraction result and the measurement diffraction result. This is especially true if a separate segmentation can be performed for the measurement diffraction result to provide further object identification.

[0034] It can be advantageous if, in a detection method according to the invention, the step of segmenting the transmission result is carried out based on homogeneous volumes. Of course, other segmentation options are also conceivable. However, the use of homogeneous volumes offers a simple and rapid starting point for performing effective segmentation to identify objects. As already explained with reference to the preliminary analysis, this can also provide an indication of the respective material for the selection of the alarm material during the simulation step.

[0035] It is also advantageous if a detection method according to the invention is carried out for at least two different identified objects in the piece of luggage. In particular, if not just a single object in a piece of luggage, but several objects have been not only detected but also identified as potentially dangerous through segmentation, the further steps of simulation, reconstruction, and comparison can advantageously be carried out for all identified alarm objects. When executed in a computer program, this is carried out in particular sequentially by the respective computer program, so that all objects are checked step by step for the respectively assigned alarm material.

[0036] It can also be advantageous if, in a detection method according to the invention, at least two different alarm materials are assigned to the at least one identified object for generating correspondingly at least two different, simulated simulation detector signals. Typically, not just a single alarm material is possible for the identified object. For example, if there is an indication of explosives, it may be useful to check for different explosive materials as alarm materials. With reference to the previous paragraph, it should be noted here that for different potential alarm objects in a piece of luggage, different alarm materials or combinations of alarm materials can naturally also be assigned for these steps of the detection method according to the invention.In addition to alarm materials, a negative check is also conceivable by assigning non-alarm materials for this simulation step and the final comparison. In other words, by actively assigning an object to a non-alarm material, the presence of an alarm material can be excluded depending on the comparison result.

[0037] A further advantage is achieved by using a classification method in a recognition method according to the invention when carrying out the comparison, in particular one of the following: correlation-based classification (in particular a one-dimensional feature vector),

[0038] Support vector machines (specifically, a multidimensional feature vector), neural networks (specifically, an abstract nonlinear (and high-dimensional) feature extraction using convolutional layers and their interpretation using fully connected layers). The above list is not exhaustive.

[0039] Of course, combinations of the individual classifications are also conceivable.

[0040] Correlation-based binary classification describes the classification of substances based on a similarity value (similarity metric). Exceeding a defined threshold leads to an alert for the material being investigated. Falling below the threshold leads to a non-alarm. An example of a similarity value could be the normalized scalar product of the signal with the respective reference.

[0041] A support vector machine (SVM) classifies spectra based on multiple features. These features (e.g., several different similarity metrics) are previously extracted from the signal to be classified. The number of features determines the dimensionality of the "feature space" (a vector space). The SVM then attempts to find an optimal hyperplane (maximum distance between support vectors) in this feature space to separate the objects that should be alerted from those that should not. The separation by the generated hyperplane can be linear (linear kernel) or nonlinear (polynomial, radial).

[0042] In a CNN (convolutional neural network), manual feature extraction is avoided. Instead, weights are trained from a large number of convolutional kernels (one-dimensional vectors or matrices). These convolutions summarize the local environment using filters (the convolution). Pooling layers ensure that the filtered data is reduced by discarding irrelevant information. The result is a high-dimensional, trained, nonlinear (nonlinear using nonlinear activation functions: Sigmoid, ReLu, etc.) feature space (the filters were trained so that the subsequent classification is as optimal as possible). The features could theoretically be used by SVMs for classification. However, fully connected neural networks are typically used to interpret the high-dimensional and nonlinear feature space and generate a classification result.The neural network thus consists of two parts: a first part for feature extraction (convolutional neural network) and a second part (fully connected neural network) which performs the separation of the high-dimensional feature space. Both parts are trained simultaneously in a learning process. It can also be advantageous if, in a recognition method according to the invention, the comparison provides a similarity value which reflects the similarity between the measured diffraction result and the simulated diffraction result. This similarity value is, in particular, an abstract value which, as a single value, is easily comparable with a threshold value in order to generate the comparison result. If such a similarity value is large, for example, a low degree of similarity can be assumed, whereas a low similarity value may indicate a high similarity.Accordingly, a single defined limit value can now be specified for the evaluation and output of the comparison result, above which the degree of similarity is sufficient to confirm a match with the assigned alarm material.

[0043] The present invention also provides a computer program product comprising instructions that, when executed by a computer, cause the computer to perform the steps of a recognition method according to the invention. Thus, a computer program product according to the invention provides the same advantages as those explained in detail with reference to a recognition method according to the invention.

[0044] Further advantages, features, and details of the invention will become apparent from the following description, which describes embodiments of the invention in detail with reference to the drawings. The features mentioned in the claims and in the description may be essential to the invention individually or in any combination. They show schematically:

[0045] Fig. 1 is a schematic representation of a method according to the prior art,

[0046] Fig. 2 shows an embodiment of a recognition method according to the invention,

[0047] Fig. 3 shows a further embodiment of a recognition method according to the invention,

[0048] Fig. 4 shows a further embodiment of a recognition method according to the invention. Figure 1 schematically illustrates how recognition has previously taken place in a recognition system 10, which is embodied here as a system-of-systems device. The recognition system 10 is equipped with a transmission device 20, for example, a computer tomography scanner. This transmission device 20 performs the transmission analysis TA on the piece of luggage G, thereby obtaining a three-dimensional transmission result TE. Using segmentation, objects 0, for example, a single object O here, can be identified. When an alarm is triggered, the piece of luggage G can then be subjected to further analysis, here using a diffraction device 30.This diffraction device 30 is designed as an X-ray diffraction system, so that diffracted X-rays are detected by a corresponding detector when passing through the piece of luggage G and output as the result of the diffraction analysis DA and thus as the measurement diffraction result MDS. The measurement detector signal MDS is back-projected by reconstruction to a measurement detector result MDE, so that a comparison can finally be made for the potential alarm object AO with an alarm database ADB and the idealized comparison values ​​stored therein. As already explained, this entails disadvantages, since actual measurement errors or impairments of the measurement detector signal MDS and / or the measurement diffraction result MDE are not taken into account in the idealized alarm database ADB.

[0049] Figure 2 shows an embodiment of the present invention for resolving the aforementioned problems. Here, too, a transmission analysis TA is performed using the transmission device 20, so that a three-dimensional transmission result TE can be provided. This is capable of identifying objects O via segmentation and, in particular, of identifying potential alarm objects AO using a preliminary analysis. For example, within the scope of such a preliminary analysis, an indication can also be given of a material group for which the identified potential alarm object AO is to be checked in the following step.

[0050] In the detection method according to the invention, a simulation detector signal SDS is generated using a diffraction simulation DS for the identified potential alarm object AO. To perform this diffraction simulation DS, not only is the subsequent diffraction analysis DA simulated, but an alarm material AM is also assumed and assigned to the alarm object AO. This results in a simulation detector signal SDS, which can be reconstructed into a simulation diffraction result SDE by backprojection. In other words, this is a simulation section that outputs a prediction and thus a projection of an expected simulation diffraction result SDE for the potential alarm object AO with the assigned alarm material AM.

[0051] The diffraction analysis DA is also carried out in the known manner so that a measurement detector signal MDS can be recorded. The measurement detector signal MDS is also reconstructed into a measurement diffraction result MDE by means of backprojection, preferably in an identical or essentially identical manner to the simulation detector signal SDS. In contrast to the known solutions, as shown, for example, in Figure 1, the comparison is now not made with an idealized alarm database ADB, but rather between the specific measurement diffraction result MDE for the alarm object AO and the simulation diffraction result SDE specific and thus individual for this piece of luggage G and this alarm object AO. Accordingly, for the same alarm material AM, for a different alarm object AO and / or a different piece of luggage G, the simulation diffraction result SDE will look different than for the present case.The simulation diffraction result SDE thus changes for each piece of luggage G and each object O, so that for each diffraction analysis DA and in the final comparison, a simulated simulation diffraction result SDE is always available that is individualized for this situation. In the final step, the comparison result VE is output, which specifically determines whether or not the alarm object AO exhibits a sufficiently high similarity to the assumed and assigned alarm material AM.

[0052] Figure 3 shows a possible further development of the embodiment of Figure 2. Here, when observing and applying the diffraction simulation DS, not only the alarm object AO itself is considered, but also neighboring areas next to the alarm object AO. Here, the three-dimensional transmission result TE is schematically represented with individual cube-shaped voxels, so that neighboring voxels next to the alarm object AO can now also be considered with the diffraction simulation DS when simulating and generating the simulation detector signal SDS. The neighboring areas next to the alarm object AO are preferably incorporated into the diffraction simulation DS using a material M that is a non-alarm material.

[0053] In the embodiment of Figure 3, the comparison result VE is further defined. Here, it is provided with a similarity value AW, which, in particular, has a scalar configuration. Such a similarity value can be a simple parameter that can be compared with a threshold value to indicate the degree of similarity and thus easily distinguish whether or not the alarm object AO is similar to the associated alarm material AM.

[0054] Figure 4 also shows several possible refinements of a detection method according to the invention. Firstly, it is clearly visible how the diffraction simulation DS is not performed just once for this alarm object AO. Rather, the assignment of different alarm materials AM, for example, takes place multiple times, so that the diffraction simulation DS is also performed multiple times. This leads to two different simulation detector signals SDS and, accordingly, two different simulation diffraction results SDE. In the final result, the one measurement diffraction result MDE is now compared multiple times with different simulation diffraction results SDE, resulting in different comparison results VE specific to each alarm material AM.Thus, a potential alarm object AO can be tested for different alarm materials AM with the same effort, since the diffraction analysis DA only needs to be carried out once and the corresponding test situation is integrated into the multiple application of the diffraction simulation DS.

[0055] The above explanation of the embodiments describes the present invention exclusively by way of examples. Individual features of the embodiments can be freely combined with one another, provided that they are technically feasible, without departing from the scope of the present invention.

[0056] 10 Detection system

[0057] 20 Transmission device

[0058] 30 Diffraction device

[0059] TA Transmission Analysis

[0060] TE transmission result

[0061] DA diffraction analysis

[0062] MDE measurement diffraction result

[0063] SDE simulation diffraction result

[0064] VE comparison result

[0065] AW similarity value

[0066] DS diffraction simulation

[0067] SDS simulation detector signal

[0068] MDS measurement detector signal

[0069] M Material

[0070] AM alarm material

[0071] 0 object

[0072] AO alarm object

[0073] G piece of luggage

[0074] ADB alarm database

Claims

Patent claims 1 . Detection method for detecting alarm objects (AO) in luggage (G), comprising the following steps: - Carrying out a transmission analysis (TA) on a piece of luggage (G) to generate a transmission result (TE), - Segmenting the transmission result (TE) to identify objects (0), - generating a simulated simulation detector signal (SDS) for at least one identified object (0) with assignment of an alarm material (AM) by means of a diffraction simulation (DS) specific for a diffraction analysis (DA), - Carrying out a diffraction analysis (DA) on the same piece of luggage (G) to generate a measurement detector signal (MDS), - Reconstruction of a measurement diffraction result (MDE) from the generated measurement detector signal (MDS) for the at least one object (0), - Reconstruction of a simulation diffraction result (SDE) with the same dimensionality as the measurement diffraction result (MDE) on the basis of the generated simulation detector signal (SDS) for the at least one object (0), - Comparison of the measurement diffraction result (MDE) with the simulation diffraction result (SDE), - Output of a comparison result (VE).

2. Recognition method according to claim 1, characterized in that a simulated simulation detector signal (SDS) is additionally generated for adjacent areas next to the at least one identified object (0) with an associated material (M).

3. Recognition method according to one of the preceding claims, characterized in that a preliminary analysis is carried out for at least one identified object (0), in particular with regard to at least one of the following parameters: - Shape of the at least one identified object (0) - Density of the at least one identified object (0) 4. Recognition method according to claim 3, characterized in that the further analysis is carried out depending on the result of the preliminary analysis.

5. Detection method according to one of the preceding claims, characterized in that the diffraction simulation (DS) takes into account at least one of the following parameters: - Angular resolution of the diffraction analysis (DA) to be performed - spatial resolution of the diffraction analysis to be performed, which can lead to signal mixing, - Attenuation of the scattered signal of the diffraction analysis (DA) to be performed by other objects (0) in the luggage (G) and the associated expected distribution of the variance of the (Poisson) noise over the momentum transfer axis (energy axis) - Attenuation of the detector response during the diffraction analysis (DA) to be performed 6. Recognition method according to one of the preceding claims, characterized in that a backprojection or reconstruction is carried out to generate the measurement diffraction result (MDE) and / or the simulation diffraction result (SDE).

7. Recognition method according to one of the preceding claims, characterized in that the step of segmenting the transmission result (TE) is carried out on the basis of homogeneous volumes.

8. Recognition method according to one of the preceding claims, characterized in that it is carried out for at least two different identified objects (0) of the piece of luggage (G).

9. Detection method according to one of the preceding claims, characterized in that for the at least one identified object (0) at least two different alarm materials (AM) are assigned for generating correspondingly at least two simulated simulation detector signals (SDS).

10. Recognition method according to one of the preceding claims, characterized in that a classification method is used when carrying out the comparison, in particular one of the following: - Correlation-based classification - Support Vector Machines - Neural network 11 . Recognition method according to one of the preceding claims, characterized in that the comparison provides a similarity value (AW) which represents the similarity between the measurement diffraction result (MDE) and the simulation diffraction result (SDE).

12. A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of a recognition method having the features of one of claims 1 to 11.

Images