TWI781410B - Light fingerprint identifying system and method thereof - Google Patents

Abstract

A light fingerprint identifying system includes a light sensing unit, a controlling unit and a processing unit. The light sensing unit detects a reference light source and the to-be-measured light source and generates a reference digital signal and a to-be-measured digital signal, respectively. The controlling unit is electrically connected to the light sensing unit and receives the reference digital signal and the to-be-measured digital signal. The controlling unit converts the reference digital signal and the to-be-measured digital signal into a reference frequency domain light fingerprint and a to-be-measured frequency domain light fingerprint, respectively. The processing unit is electrically connected to the controlling unit and receives the reference frequency domain light fingerprint. The processing unit trains an identifying model according to the reference frequency domain light fingerprint and transmits the trained identifying model to the controlling unit. The controlling unit recognizes whether the to-be-measured frequency domain light fingerprint is the reference frequency domain light fingerprint according to the trained identifying model. Therefore, it is favorable for performing the identification from different light fingerprints.

Description

Light source light streak identification system and method

The present invention relates to an identification system and method thereof, and in particular to an identification system and method applied to a light pattern of a light source.

The phenomenon of light streaks (Light Fingerprints) of lighting equipment comes from the slight differences between the electronic components used in it, whether it is a traditional fluorescent tube or a light-emitting diode (Light-Emitting Diode) , LED), even when the brand and model are the same, the light emitted by each lighting device is slightly different from other devices. With the advancement of electronic technology and the vigorous development of artificial intelligence (AI), the application of light patterns in lighting equipment has become a hot research topic in recent years, and most light pattern technologies are only used in indoor positioning. Deep Learning classifies light fixtures in a room.

Door lock systems range from simple password locks, radio frequency identification (RFID) locks to face or fingerprint biometric devices, etc., usually need to carry or install additional hardware and software equipment (such as: RFID cards), or even There is a risk of being stolen and copied, and the cost of biometric devices is high.

In order to solve the above problems, the present invention provides a light source light pattern identification system and its method, which can directly identify and identify through the light pattern of the light source, and use a low-cost processor to realize artificial intelligence to complete the function of the access control system.

According to an embodiment of a structural aspect of the present invention, a light source light pattern recognition system is provided, which includes a light sensing unit, a control unit, and a processing unit. The light sensing unit detects the reference light source and the test light source, and generates a reference digital signal and a test digital signal respectively. The control unit is electrically connected to the light sensing unit and receives the reference digital signal and the digital signal to be measured, and the control unit respectively converts the reference digital signal and the digital signal to be measured into a reference frequency domain light pattern and a frequency domain light pattern to be measured. pattern. The processing unit is electrically connected to the control unit and receives the reference frequency domain light pattern, the processing unit trains a recognition model according to the reference frequency domain light pattern, and sends the trained recognition model to the control unit. Wherein, the control unit identifies whether the frequency-domain light pattern to be measured is a reference frequency-domain light pattern according to the trained identification model.

In this way, the corresponding identification model can be trained by detecting the reference light source through the light sensing unit, and the trained identification model can be used to identify and identify the light source to be tested, so as to achieve low system cost and simple implementation process.

According to the light source light pattern identification system of the above-mentioned embodiment, when the frequency domain light pattern to be measured is the reference frequency domain light pattern, the control unit generates a control signal; when the frequency domain light pattern to be measured is not the reference frequency domain light pattern , the control unit generates another control signal.

According to the light source light pattern identification system of the above-mentioned embodiment, wherein the light sensing unit includes a light sensor and an amplifier. The light sensor detects the reference light source and the light source under test, and generates a reference initial signal and an initial signal under test respectively. The amplifier is electrically connected to the light sensor, and converts the reference initial signal and the initial signal to be tested into a reference digital signal and a digital signal to be tested respectively.

According to the light source light streak identification system of the embodiment described in the preceding paragraph, wherein After the recognition model is trained, the processing unit converts the trained recognition model into a language code, so that the trained recognition model is constructed in the control unit.

According to the light source light pattern identification system of the above-mentioned embodiment, the control unit transmits the reference frequency domain light pattern from the control unit to the processing unit through a transmission interface, and transmits the trained identification model from the processing unit to the control unit.

According to an embodiment of a method aspect of the present invention, a light source light pattern identification method is provided, which includes a sensing step, a sampling step, a training step and an identifying step. The sensing step is to provide a light sensing unit to sense a reference light source and a test light source, and generate a reference digital signal and a test digital signal respectively. The sampling step is to use a control unit to sample the reference digital signal and the digital signal to be measured, and respectively transform the reference digital signal and the digital signal to be measured into a reference frequency domain light pattern and a measured frequency domain light pattern through Fourier transform. The training step is to provide a processing unit to receive the reference frequency domain light pattern, and the processing unit trains a recognition model according to the reference frequency domain light pattern, and sends the trained recognition model to the control unit. The identifying step is to drive the control unit to identify whether the frequency-domain light pattern to be measured is the reference frequency-domain light pattern according to the trained identification model.

In this way, through the artificial intelligence identification training method of the processing unit, the identification model can be completed and then transmitted to the control unit as an identification basis, and another reference frequency domain light pattern can be added to train another identification model to achieve the best use convenience.

The light source light streak identification method according to the implementation manner mentioned in the preceding paragraph further includes a control step. The control step is to drive the control unit when the frequency domain light pattern to be measured is the reference frequency domain light pattern, the control unit generates a control signal; the control step is to drive the control unit when the frequency domain light pattern to be measured is not the reference frequency domain light pattern, then The control unit generates another control signal.

According to the light source light pattern identification method of the implementation manner mentioned in the preceding paragraph, the sensing step includes a detection step and an amplification step. The detection step is to detect the reference light source and the test light source through a light sensor, and generate a reference initial signal and a test initial signal respectively. The amplifying step is to respectively amplify the reference initial signal and the initial signal to be tested into a reference digital signal and a digital signal to be tested through an amplifier.

According to the light source light pattern identification method of the implementation manner mentioned in the preceding paragraph, the training step includes a establishing step and a correcting step. The establishment step is to establish and store the reference frequency domain light pattern in the processing unit. The correction step is to drive the processing unit to modify and optimize the complex parameters of the identification model according to the reference frequency domain light pattern, so as to generate the trained identification model.

According to the method for identifying light streaks of light sources in the aforementioned implementation manner, the steps are performed in a sequence of sensing, sampling, training, identifying, and controlling.

Several embodiments of the present invention will be described below with reference to the drawings. For the sake of clarity, many practical details are included in the following narrative. It should be understood, however, that these practical details should not be used to limit the invention. That is, in some embodiments of the present invention, these practical details are unnecessary. In addition, for the sake of simplifying the drawings, some commonly used structures and elements will be shown in a simple and schematic way in the drawings; and repeated elements may be denoted by the same reference numerals.

FIG. 1 is a schematic diagram of a light source light pattern identification system 100 according to an embodiment of a structural aspect of the present invention. It can be seen from FIG. 1 that the light source light pattern recognition system 100 includes a light sensing unit 110 , a control unit 120 and a processing unit 130 . The light sensing unit 110 detects the reference light source 102 and the test light source 104 to generate a reference digital signal 112 and a test digital signal 114 respectively, wherein each light source can be any lighting device, such as a backlight module of a mobile phone. The control unit 120 is electrically connected to the light sensing unit 110 and receives the reference digital signal 112 and the digital signal to be measured 114 output from the light sensing unit 110, and the control unit 120 respectively passes the reference digital signal 112 and the digital signal to be measured 114 through The Fourier transform is converted into a reference frequency-domain light pattern 121 and a frequency-domain light pattern to be measured (not shown). The processing unit 130 is electrically connected to the control unit 120 and receives the reference frequency domain light pattern 121 . The processing unit 130 trains a recognition model 131 according to the reference frequency domain light pattern 121 , and sends the trained recognition model 131 back to the control unit 120 . Next, the control unit 120 identifies whether the frequency-domain optical pattern to be tested is the same as the reference frequency-domain optical pattern 121 according to the trained identification model 131 .

In this way, the light source light pattern identification system 100 of the present invention can detect the reference light source 102 through the light sensing unit 110 to train the corresponding identification model 131, and then use the trained identification model 131 to identify the light source 104 to be tested, which can be used Low system cost achieves perfect identification function.

FIG. 2 is a schematic diagram of a light source light pattern identification system 100 according to an embodiment of another structural aspect of the present invention. It can be seen from FIG. 2 that the light sensing unit 110 may include a light sensor 110a and an amplifier 110b. The light sensor 110a detects the reference light source 102 and the light source under test 104 , and generate a reference initial signal 1021 and a test initial signal 1041 respectively. The amplifier 110b is electrically connected to the light sensor 110a, and converts the reference initial signal 1021 and the under-test initial signal 1041 into the reference digital signal 112 and the under-test digital signal 114 respectively. In detail, the control unit 120 can be a microcontroller (Microcontroller Unit, MCU), and the light sensor 110a can be a photodiode (Photodiode, PD), wherein the sampling voltage range of the MCU is mostly 0~3.3V or 5V. Therefore, the reference initial signal 1021 and the initial signal to be tested 1041 are amplified by the amplifier 110 b to remove high-frequency noise, thereby improving the resolution of sampling.

It should be noted that the amplifier 110b may include an analog-to-digital converter (Analog-to-Digital Converter, ADC). The reference initial signal 1021 and the initial signal to be tested 1041 are converted from analog signals to digital signals through the ADC. In addition, the reference frequency domain light pattern 121 can be transmitted from the control unit 120 to the processing unit 130 through a transmission interface (not shown) between the control unit 120 and the processing unit 130, and the trained identification model 131 can be transmitted from the processing unit 130 is transmitted to the control unit 120, and the transmission interface is a known technology and is not the focus of the present invention, and the details will not be repeated.

As can be seen from FIG. 2 , the light source light pattern identification system 100 may further include a switch 140 . The switch 140 is electrically connected to the control unit 120 . The control unit 120 controls the switch 140 through a general-purpose input/output (GPIO), so as to activate the switch 140 . Specifically, when the frequency-domain light pattern to be measured is the reference frequency-domain light pattern 121, the control unit 120 generates a control signal (not shown) and controls the switch 140 to turn on; if the frequency-domain light pattern to be measured is not When referring to the frequency-domain light pattern 121, the control unit 120 generates another control signal (not shown) and controls the switch 140 to be closed, wherein the switch 140 can be an anti-theft lock, but not limited thereto.

FIG. 3 is a flow chart showing the steps of the method S100 for identifying light streaks of a light source according to an embodiment of a method aspect of the present invention. It can be seen from FIG. 3 that the light source light pattern identification method S100 includes a sensing step S110 , a sampling step S120 , a training step S130 and an identification step S140 .

The sensing step S110 is to provide the light sensing unit 110 to sense the reference light source 102 and the test light source 104 , and the light sensing unit 110 generates a reference digital signal 112 and a test digital signal 114 respectively. The sampling step S120 is to use the control unit 120 to sample the reference digital signal 112 and the digital signal to be measured 114, and respectively transform the reference digital signal 112 and the digital signal to be measured 114 into the reference frequency domain light pattern 121 and the frequency domain to be measured light pattern . In the training step S130 , the processing unit 130 receives the reference frequency domain light pattern 121 , and the processing unit 130 trains the identification model 131 according to the reference frequency domain light pattern 121 , and sends the trained identification model 131 to the control unit 120 . In the identifying step S140 , the drive control unit 120 identifies whether the frequency-domain optical pattern to be measured is the reference frequency-domain optical pattern 121 according to the trained identification model 131 and generates an identification result (not shown separately).

In this way, through the artificial intelligence identification training method of the processing unit 130, the identification model 131 can be completed and then transmitted to the control unit 120 as an identification basis, and if an additional identification object (i.e. light source) is to be added, only the identification model 131 needs to be retrained , or add another recognition model 131 and construct it in the control unit 120, so as to achieve convenience in use.

In addition, as can be seen from FIG. 3 , the method S100 for identifying light streaks of a light source may further include a control step S150 . In the control step S150 , the drive control unit 120 controls the switch 140 according to the identification result of the identification step S140 (whether the frequency-domain light pattern to be tested is the reference frequency-domain light pattern 121 ). Wherein, when the frequency domain light pattern to be measured is the reference frequency domain light pattern 121, the control unit 120 generates a control signal and controls the switch 140 to open; when the frequency domain light pattern to be measured is not the reference frequency domain light pattern 121, then controls The unit 120 generates another control signal and controls the switch 140 to close. Furthermore, the execution order of the steps of the light source pattern identification method S100 may be a sensing step S110 , a sampling step S120 , a training step S130 , an identification step S140 and a control step S150 . Thus, the light source light pattern identification method S100 of the present invention can be used for identity identification based on the light pattern of each light source and applied to anti-theft locks, electric locks or opening and closing devices, but not limited thereto.

FIG. 4 is a flowchart showing the steps of the sensing step S110 of the light source light streak identification method S100 according to the embodiment of the aspect in FIG. 3 . It can be seen from FIG. 4 that the sensing step S110 may include a detection step S111 and an amplification step S112. The detection step S111 is to detect the reference light source 102 and the test light source 104 through the light sensor 110a, and generate a reference initial signal 1021 and a test initial signal 1041 respectively. The amplifying step S112 is to respectively amplify the reference initial signal 1021 and the under-test initial signal 1041 through the amplifier 110 b into the reference digital signal 112 and the under-test digital signal 114 . In detail, after the light source is converted by the light sensor 110a, the obtained light source signal (ie, the reference initial signal 1021 and the initial signal to be tested 1041) is about 300mV to 500mV, and the light source signal is then passed through the amplifier 110b to increase the gain, so as to make The control unit 120 performs signal sampling.

FIG. 5 is a flowchart showing the steps of the training step S130 of the light source light streak identification method S100 according to the implementation manner of the aspect in FIG. 3 . It can be seen from FIG. 5 that the training step S130 may include a building step S131 and a modifying step S132. The establishment step S131 is to establish and store the reference frequency-domain light pattern 121 in a memory (not shown) in the processing unit 130 . The modification step S132 is to drive the processing unit 130 to modify and optimize the complex parameters of the identification model 131 according to the reference frequency domain light pattern 121 , so as to generate the trained identification model 131 . Specifically, in the training step S130 , the processing unit 130 trains the identification model 131 through an artificial intelligence algorithm, and the algorithm is a known technology in the technical field, and will not be repeated here.

In addition, the establishment step S131 is to process and store the reference frequency domain light pattern 121 in conjunction with the signal sampling of the control unit 120, and the correction step S132 is to use a grid search method (Grid Search, GS) and a cross-validation (Cross Validation (CV) optimizes the parameters of the identification model 131, where the parameters may include the number of nodes in each layer, the number of hidden layers, sampling points, training times, and identification thresholds. In addition, when the recognition model 131 is trained, the processing unit 130 converts the trained recognition model 131 into a C language code based on the One-vs.-Rest (OVR) method, so that the trained recognition model 131 is constructed in the control unit 120.

Based on the above, the present invention has the following advantages: First, the identity can be identified based on the different light patterns of different light source devices. Second, the optimized identification model can be used as the identification basis through the artificial intelligence identification training method of the processing unit. Third, a lower cost system can be used to perform light pattern identification for identity or device screening.

Although the present invention has been disclosed above in terms of implementation, it is not intended to limit the present invention. Anyone skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection of the present invention The scope shall be defined by the scope of the appended patent application.

100: Light source light pattern identification system 102: Reference light source 104: The light source to be tested 1021: Refer to the initial signal 1041: Initial signal to be tested 110: Light sensing unit 110a: light sensor 110b: Amplifier 112: Reference digital signal 114: digital signal to be tested 120: Control unit 121: Reference frequency domain light pattern 130: processing unit 131: Identification model 140: switch S100: Light source light pattern identification method S110: Sensing step S111: detection step S112: enlargement step S120: sampling step S130: training steps S131: Establishment steps S132: Correction steps S140: identification step S150: Control steps

Figure 1 is a schematic diagram of a light source light pattern recognition system according to an embodiment of a structural aspect of the present invention; Figure 2 is a schematic diagram of a light source light pattern recognition system according to an embodiment of another structural aspect of the present invention; Fig. 3 is a flow chart showing the steps of a light source light pattern identification method according to an embodiment of a method aspect of the present invention; FIG. 4 is a flow chart showing the sensing steps of the light source light streak identification method according to the embodiment of the aspect in FIG. 3; and FIG. 5 is a flow chart showing the training steps of the light source light pattern identification method according to the embodiment of the embodiment in FIG. 3 .

S100: Light source light pattern identification method

S110: Sensing step

S120: sampling step

S130: training steps

S140: identification step

S150: Control steps

Claims (8)

A light source light pattern identification system, comprising: a light sensing unit, which detects a reference light source and a test light source, and generates a reference digital signal and a test digital signal respectively; a control unit, which is electrically connected to the light sensor The measuring unit receives the reference digital signal and the digital signal to be measured, and the control unit respectively transforms the reference digital signal and the digital signal to be measured into a reference frequency domain light pattern and a frequency domain light pattern to be measured; and A processing unit, electrically connected to the control unit and receiving the reference frequency domain light pattern, the processing unit trains a recognition model according to the reference frequency domain light pattern, and sends the trained recognition model to the control unit; wherein, The control unit identifies whether the light pattern in the frequency domain to be measured is the light pattern in the reference frequency domain according to the trained identification model; wherein, when the light pattern in the frequency domain to be measured is the light pattern in the reference frequency domain, the control unit generates A control signal controls a switch to be turned on; wherein, when the frequency-domain light pattern to be measured is not the reference frequency-domain light pattern, the control unit generates another control signal and controls the switch to be turned off. The light source light pattern identification system as described in Claim 1, wherein the light sensing unit includes: a light sensor that detects the reference light source and the light source to be tested, and generates a reference initial signal and a test initial signal respectively signal; and an amplifier, electrically connected to the light sensor, and respectively converting the reference initial signal and the initial signal to be tested into the reference digital signal and the digital to be measured signal. The light source light pattern recognition system as described in Claim 1, wherein, after the recognition model is trained, the processing unit converts the trained recognition model into a language code, so that the trained recognition model is constructed on the control unit. The light source light pattern identification system as described in Claim 1, wherein the control unit transmits the reference frequency domain light pattern from the control unit to the processing unit through a transmission interface, and transmits the trained identification model from the processing unit transmitted to the control unit. A light source light pattern identification method, comprising: a sensing step, which is to provide a light sensing unit to sense a reference light source and a test light source, and respectively generate a reference digital signal and a test digital signal; a sampling step, A control unit is used to sample the reference digital signal and the digital signal to be measured, and respectively convert the reference digital signal and the digital signal to be measured into a reference frequency domain light pattern and a frequency domain light pattern to be measured ; a training step, providing a processing unit to receive the reference frequency domain light pattern, and the processing unit trains a recognition model according to the reference frequency domain light pattern, and sends the trained recognition model to the control unit; The step is to drive the control unit according to the trained recognition model Identifying whether the optical pattern in the frequency domain to be measured is the reference frequency domain optical pattern; and a control step, comprising: driving the control unit when the optical pattern in the frequency domain to be measured is the reference frequency domain optical pattern, then the control unit Generate a control signal and control a switch to turn on; and drive the control unit when the frequency-domain optical pattern to be measured is not the reference frequency-domain light pattern, then the control unit generates another control signal and controls the switch to close. The light source light pattern identification method as described in Claim 5, wherein the sensing step includes: a detection step, which is to detect the reference light source and the test light source through a light sensor, and generate a reference initial signal respectively and an initial signal to be tested; and an amplifying step, which is to respectively amplify the reference initial signal and the initial signal to be tested into the reference digital signal and the digital signal to be tested through an amplifier. The light source light pattern identification method as described in claim 5, wherein the training step includes: a building step, which is to create and store the reference frequency domain light pattern in the processing unit; and a correction step, which is to drive the processing unit according to the The complex parameters of the identification model are corrected and optimized with reference to the frequency-domain light pattern, so as to generate the trained identification model. The light source light pattern identification method as described in Claim 5, wherein the sequence of performing the steps is the sensing step, the sampling step, the training step, the identification step and the control step.

Images