A high-efficiency acquisition method of LED-multispectral images based on frequency-division modulation and RGB camera
Introduction
Multispectral imaging is typically used to capture spectral reflectance of the imaged scene. And it is not limited to the visual range, but also can be applied in ultraviolet, near infrared, and infrared spectrum as well, depending on the response range of the sensor. Through the spectral imaging systems, the spectral reflectance of a scene reflecting the unique property of an object can be recovered from the acquired images. Therefore, multispectral imaging has a wide range of applications, such as remote sensing [1], biometric security [2], [3], medical imaging [4], [5], culture and heritage [6] etc.
There are already many different types of multispectral imaging techniques and systems. Among them, one promising technique of multispectral imaging, which is of the primary interest in this paper, is based on Light Emitting Diode (LED) illumination [7], [8], [9], [10], [11]. In recent years, LED illumination based multispectral imaging (LEDMSI) has attracted much attention due to its rapid computer-controlled switching capability, robustness, and cost-effectiveness. Availability of high-intensity LEDs with many different colors and peak wavelengths spanning the whole visual range and even infrared region makes it possible to construct more effective multispectral systems. LEDMSI has been used in several applications like underwater geological and biological surveys [12], medical imaging [13] and film scanner [14]. The systems can be classified into two major categories according to the type of camera: monochrome camera based [7], [8], and RGB camera based [10], [11]. In a monochrome camera based LEDMSI (Mono-LEDMSI) system, a group of n different types of LEDs are selected, each type of LED is illuminated in a sequence, and a monochrome camera captures images under the illuminated LED, thereby generating n-band multispectral images. In an RGB camera based LEDMSI (RGB-LEDMSI) system, combinations of LEDs whose number and wavelength might be constrained by the availability and other practical factors of the system are selected. Each combination of LEDs is lit at a time, and a color camera is used to capture 3-band images. A K 3 N-band image of a scene can be acquired with N different combinations of LEDs in N exposures, so the acquisition can be speed up by a factor of three.
For monochrome camera and frequency-division modulation, our research team has proposed a number of imaging methods. Frequency-division modulation can not only suppress interference, but also can achieve the acquisition of multi-wavelength images simultaneously, saving image acquisition time. In 2018, X. Yang et al. proposed an optimized illumination method, in which a shaped-function signal was applied to increase the dynamic range of LED-multispectral imaging system [15]. In 2019, H. Li et al. proposed and demonstrated the multi-wavelength “synergy effect” in LED-multispectral images obtained by frequency-division modulation [7]. In the same year, F. Liu et al. proposed a method combining modulation–demodulation–frame-accumulation technique and pattern recognition to realize heterogeneity classification [16], in which the modulation–demodulation–frame-accumulation technique was adopted in pre-processing experiment to enhance the image signal. In the above research, a monochrome camera was used as the image acquisition terminal, and the modulation and demodulation were reasonably used in the process to achieve different application purposes.
In addition, for RGB camera and multiplexed illumination (a combination of two or more LEDs in each exposure), some researchers have proposed their systems. Rather than sequentially activating the sources, J. Park et al. proposed a novel technique to find the optimal multiplexing sequence of spectral sources in 2004 [10], so as to minimize the number of acquired images. In 2012, M. Parmara et al. developed a multispectral imaging system that captures multiple acquisitions during a rapid sequence of LED with different colors [17]. In 2015, R. Shrestha et al. proposed a fast multispectral imaging using LED illumination and an RGB camera along with a novel LED selection method [18]. In the same year, R. Shrestha et al. proposed a system which used an RGB camera along with two or three combinations of three different types of LEDs so as to acquire multispectral images of six or nine channels [11]. In 2018, T. Heimpold proposed a new time efficient selection method for the design process of an illumination. It chose an optimal LED combination from an existing database to match a predefined spectral power distribution [19]. From the above, we can get that the researchers are particularly concerned about finding the optimal multiplexed illumination of LEDs in RGB-LEDMSI systems. The determined multiplexed illuminations are activated sequentially, and the corresponding images are collected, then a multispectral video of the scene is obtained.
In the past research work, some have combined frequency-division modulation to improve image quality and increase the dynamic range of LED-multispectral imaging system, and some have proposed multiplexed illuminations in RGB-LEDMSI to speed up image acquisition, but no one has tried to combine the two. When the two are combined, the entire image acquisition process is equivalent to be performed on two dimensions: (1) the color channel of the receiving device itself; (2) frequency-division modulation of multi-wavelength images. Therefore, a high-efficiency method for obtaining multispectral images can be proposed, which may provide a reference for improving the quality of tissue multispectral transmission imaging in the future.
Section snippets
Frequency-division modulation technology
(a) Principle of frequency-division modulation technology
Grayscale level is of vital importance for image accuracy, and the more grayscale levels the image contains, the greater the amount of image information. Through modulation–demodulation technique or frame accumulation technique, the grayscale resolution of images can be increased to some extent [20], [21].
Frequency-division modulation is a technique by which the total bandwidth is divided into a series of non-overlapping frequency bands,
Experiment device
The concept underlying our LED-multispectral imaging system is shown in Fig. 3. The system consists of the five parts: light source part, control part, image acquisition part, processing part and other parts. The light source part mainly includes: LED panel (including 6 types of LEDs with 420 nm, 460 nm, 520 nm, 570 nm, 620 nm and 690 nm () as the center wavelengths) and multifunctional signal generator (model: AFG-2105). The control part includes LED driving circuit of PT4115
Results and discussion
By combining multi-channel transmission of frequency-division modulation and the color splitting of RGB camera, a high-efficiency acquisition method of LED-multispectral images is implemented. Further, we evaluated the quality of the multispectral images obtained by the method proposed in this paper. Image quality can be assessed using subjective or objective methods. Subjective IQAs are a method based on the way in which humans experience or perceive image quality. In the objective method, IQA
Conclusion
In this paper, we propose a high-efficiency acquisition method of LED-multispectral images based on frequency-division modulation and RGB camera. The acquisition of LED-multispectral images is considered from the two dimensions of the RGB camera’s color channel and multi-wavelength frequency-division modulation. Frequency-division modulation technology can achieve simultaneous acquisition of multi-wavelength images by setting different carrier frequencies. RGB camera uses its own color filters
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgment
We thank the State Key Laboratory of Precision Measuring Technology and Instruments for the use of their equipment.
References (23)
- et al.
Multi-spectral lowrank structured dictionary learning for face recognition
Pattern Recognit.
(2016) - et al.
Synergy effect and its application in LED-multispectral imaging for improving image quality
Opt. Commun.
(2019) - et al.
Employment of the appropriate range of sawtooth-shaped-function illumination intensity to improve the image quality
Optik
(2018) - et al.
Land cover classification of multispectral remote sensing images based on time-spectrum association features and multikernel boosting incremental learning
J. Appl. Sens.
(2019) - et al.
Multispectral palmprint recognition using multiclass projection extreme learning machine and digital shearlet transform
Neural Comput. Appl.
(2016) - et al.
Laser-based optical spectroscopy device for breast density and breast cancer risk pre-screening
J. Biophotonics
(2017) - et al.
Use of hyperspectral/multispectral imaging in gastroenterology. shedding some-different-light into the dark
J. Clin. Med.
(2019) - C. Colantonio, C. Pelosi L. D’Alessandro, S. Sottile, G. Calabro, M. Melis, Hypercolorimetric multispectral imaging...
- et al.
Camera system for multispectral imaging of documents
Proc. SPIE
(2009) - et al.
Fuzzy evaluation output of taste information for liquor using electronic tongue based on cloud model
Sensors
(2020)
Cited by (12)
Modulation and demodulation with quadrature square waves as carriers to improve spectrum utilization
2023, Digital Signal Processing: A Review JournalAn underwater image restoration based on global polarization effects of underwater scene
2023, Optics and Lasers in EngineeringThreshold method to improve the speed of “Registration-Frame accumulation” and its application in breast multispectral transmission images
2023, Infrared Physics and TechnologyA modulation method that can improve the performance of LED multi-spectral imaging
2023, Spectrochimica Acta - Part A: Molecular and Biomolecular SpectroscopyCitation Excerpt :This limits the selection of LED band and causes color channel crosstalk. Frequency division multiplexing can realize efficient signal transmission by shifting the signal spectrum to different frequency bands, so many researchers applied it into LEDMSI [12–15]. However, they all used sinusoidal wave as the carrier signal.
LED multispectral imaging based on frequency-division modulation of square wave and synchronous triggering
2022, OptikCitation Excerpt :Frequency-division modulation-frame accumulation technology is used to enhance image signals in preprocessing experiments. Li et al. [28]. proposed an efficient LEDMSI based on the combination of frequency-division modulation and RGB camera, which further improves the acquisition speed of multispectral images on the premise of ensuring the quality.
Defect Detection Method of Overhead Line Pins Based on Multi-Sensor Data Acquisition of UAV
2024, Lecture Notes of the Institute for Computer Sciences, Social-Informatics and Telecommunications Engineering, LNICST