Color image sensor based on two-dimensional semiconductor heterostructure and application of color image sensor in image convolution processing

Abstract

The invention discloses a color image sensor based on a two-dimensional semiconductor heterostructure and application of the color image sensor in image convolution processing. The device comprises a bipolar two-dimensional semiconductor and an N-type two-dimensional semiconductor which are deposited on a substrate, and the bipolar two-dimensional semiconductor and the N-type two-dimensional semiconductor form a horizontal heterojunction structure. By adjusting the voltage of the back gate, the energy band arrangement of the heterojunction is controlled, a spectral response function capable of continuously changing is realized, and image information under multiple wavebands can be obtained without a color filter. Meanwhile, the photocurrent response of heterojunctions is regulated and controlled through source and drain voltages, then different weights are given to output signals, a plurality of heterojunctions with the same size as a convolution kernel are connected in parallel, and the sum of read currents is mapped into a gray value, so that a convoluted image under the wave band can be obtained. And image convolution processing can be completed by respectively applying convolution operation under different wavebands. The device has excellent performance in the aspects of integrated image acquisition and convolution calculation, effectively reduces the cost of calculation resources, and remarkably improves the image processing efficiency.

Description

Technical Field

[0001] The present invention belongs to the technical field of novel semiconductor devices and relates to a color image sensor based on a two-dimensional semiconductor heterostructure and its application in image convolution processing. Background Art

[0002] Image information usually contains multiple dimensions, such as light intensity, spatial position, color, etc. Collecting and processing high-dimensional image information helps to apply machine vision in daily life and production, and improve work efficiency. Among them, light intensity and spatial information are the most basic, usually obtained using CCD or CMOS arrays with readout circuits. With the expansion of machine vision application scenarios, more complex application scenarios and tasks usually rely on the collection and processing of color information.

[0003] In general, there are still problems in the current machine vision system for color image recognition and processing: traditional CCD or CMOS needs to be equipped with a complex structure of color filter array, which increases the device size and process complexity; the spectral response curve of traditional image sensors is fixed and cannot be modulated as needed; the modules for processing and collecting information in traditional color image recognition systems are separated, which means that additional computing equipment needs to be introduced to complete the color image processing process, such as the commonly used convolution calculation, which increases the computing resource overhead and processing time. Summary of the invention

[0004] The purpose of the present invention is to design a new type of color image sensor to solve the three problems existing in the traditional machine vision system for processing color images, namely, the fixed perception wavelength range, the overly complex sensor structure, and the separation of the devices for perception and algorithm implementation.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] According to the first aspect of the present specification, a color image sensor based on a two-dimensional semiconductor heterostructure is provided, the device comprising a bipolar two-dimensional semiconductor layer and an N-type two-dimensional semiconductor layer deposited on a substrate, the two forming a horizontal heterojunction, and the color image sensor can simultaneously realize the acquisition and processing of image data with convolution calculation through a combination of gate voltage and source-drain voltage.

[0007] Furthermore, the substrate is a P-type heavily doped silicon wafer on which silicon oxide is deposited.

[0008] Furthermore, the material of the bipolar two-dimensional semiconductor layer is molybdenum telluride.

[0009] Furthermore, the material of the N-type two-dimensional semiconductor layer is molybdenum sulfide.

[0010] Furthermore, the energy band arrangement of the heterojunction is controlled by adjusting the gate voltage, thereby realizing different spectral perception functions and collecting multi-channel image information without the need for color filters.

[0011] Furthermore, the photocurrent response intensity of the heterojunction is controlled by adjusting the source-drain voltage, thereby giving different channel weights to light signals of different wavelengths, thereby realizing the convolution calculation process of the image.

[0012] According to a second aspect of the present specification, there is provided an application of the above-mentioned color image sensor based on the two-dimensional semiconductor heterostructure in image convolution processing, specifically including:

[0013] Determine the range of optical signal intensity and channel type, and determine the mapping relationship between optical signal intensity and grayscale value;

[0014] Determine the channel type according to the color that needs to be paid attention to, and determine the gate voltage of the device according to the channel type;

[0015] Determine the convolution kernel corresponding to each channel according to the purpose of image convolution processing, calculate the weight distribution of the convolution kernel in each band and normalize it;

[0016] Establish the mapping relationship between device photocurrent and light intensity under the specified channel, as well as the mapping relationship between weight and source-drain voltage;

[0017] Arrange several identical devices according to the size and shape of the convolution kernel and connect them in parallel to form a photosensitive array;

[0018] Determine the source-drain voltage matrix of the photosensitive array according to the convolution kernel weight distribution and mapping relationship;

[0019] When collecting images, a predetermined matrix of gate voltage and source-drain voltage under different channels is applied to the photosensitive array to read the photocurrent and;

[0020] The photocurrent and photocurrent are converted into grayscale values ​​to obtain images after convolution processing under different channels, and then fused to obtain a color image.

[0021] Furthermore, the mapping relationship between the light signal intensity and the grayscale value is determined by pre-calibrating the dynamic range of the color image sensor to obtain the upper limit P of the light signal intensity that can be detected. max and the lower limit P min The expected grayscale value range of the image is recorded as G; the mapping relationship between the light signal intensity and the grayscale value is established according to the following formula: P = qG, where P is P max With P min The difference is , q is the proportional coefficient.

[0022] Furthermore, the gate voltage is determined specifically as follows:

[0023] According to the color that needs to be paid attention to, a total of c channel types are determined, denoted as C1, C2, ..., C i ,…,C c , i=1,2,…,c, where C i Represents the band corresponding to channel i;

[0024] When using C1~C c The transfer curve of the device is tested under the condition that the device is illuminated by a light signal in the wavelength range, and the corresponding gate voltage V is extracted. g1 ,V g2 ,…,V gi ,…,V gc , i=1,2,…,c, where V gi This means that at this gate voltage, the device is i The responsiveness is the highest.

[0025] Furthermore, for channel C i , apply gate voltage V to the device gi , using C i The optical signal of the wavelength band illuminates the device and scans the source-drain voltage V ds To test the output curve of the device and extract the photocurrent I ph Output curve, V ds The selection basis is to enable the device to exhibit complete rectification characteristics, and its range is set to (V ds，min ,V ds，max ), in P max With P min Internal adjustment C i The intensity of the light signal in the band is used to obtain multiple groups of photocurrents I ph The output curve is fitted using the following formula:

[0026] I ph =α(kP) β

[0027] Where P is the light signal intensity, α is the proportionality coefficient, and k is the ds function, which is related to V ds is positively correlated, and its range of variation represents V ds The ability to adjust the output signal when the light intensity is fixed, while β reflects the linear relationship between the photocurrent and the input light intensity. i In the case of band optical signals, α and β are fixed values. Repeat this step to obtain the value of each C i The corresponding α and β values;

[0028] Among them, V ds，max The corresponding maximum k value is recorded as k max , while V ds，min The corresponding minimum k value is recorded as k min, for all values ​​of k, divide by k max To complete the normalization, the normalized k value is regarded as the weight W, and the following formula is used to fit each C i W and V under ds Relationship:

[0029]

[0030] Among them, a and b reflect the change of W with V ds The speed of change of C i In the case of band optical signals, a and b are fixed values. Repeat this step to obtain each C i The corresponding a and b values.

[0031] Furthermore, the same devices are arranged according to the pixel arrangement of the convolution kernel to form a photosensitive array, each pixel corresponds to a device, all devices are connected in parallel, and channel C i The corresponding convolution kernel K i The weight of each pixel in is normalized and then substituted into the fitted channel C i Corresponding W and V ds , the source-drain voltage required to be applied to the device at the corresponding position of the pixel is obtained, thereby obtaining each C i The source-drain voltage V ds Matrix, denoted by V ds i .

[0032] Furthermore, in order to perform convolution processing on the initial image to obtain the feature map, firstly, for the selected C1~C c Each C under the wavelength range i , apply V to the gate of each device in the photosensitive array gi , the source-drain voltage is based on the matrix V ds i to apply, read the sum of the currents of all parallel devices, and then maintain V gi The convolution kernel remains unchanged, traverses the entire initial image, reads the current sum multiple times, and converts the read current sum into C i The gray value distribution diagram obtained is the gray value distribution diagram under C i The convolutional feature map under the band, combined with C1~C c All feature maps under the wavelength range are finally convolved to obtain a multi-band color image.

[0033] Beneficial effects of the present invention: Compared with the pixel units used in the existing image sensor arrays, the present invention designs a three-terminal device, using bipolar two-dimensional semiconductor molybdenum telluride and N-type two-dimensional semiconductor molybdenum sulfide to build a horizontal heterojunction on a heavily doped silicon wafer deposited with silicon oxide. Since the thickness of the material is very small, it is easy to be regulated by the back gate. The energy band arrangement of the heterojunction is modulated by applying a voltage through the back gate to obtain the responsiveness to light signals of different wavelengths, so that light signals of different bands can be differentially perceived without using a color filter. At the same time, adjusting the source-drain voltage of each device can adjust the size of the read photocurrent, thereby scaling the photocurrent within a certain range. By connecting in parallel several sensing units with the same size as the convolution kernel, the multiplication and accumulation operations of the convolution kernel and the corresponding pixel can be completed, which makes it possible to directly implement convolution processing in multiple bands at the device end. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 Structural diagram of heterojunction device built for bipolar and N-type two-dimensional semiconductors;

[0035] Figure 2 A flowchart of image convolution processing using the heterojunction device;

[0036] Figure 3 is the normalized responsivity distribution of the 450-700nm optical signal corresponding to the device when the gate voltage changes in the range of -50V to 50V;

[0037] Figure 4 This is an example of the perceived effect of the blue letter "I" and the red letter "C" at the corresponding gate voltage;

[0038] Figure 5 is the V of the blue channel and the red channel calculated according to the convolution kernel ds Matrix distribution;

[0039] Figure 6 To use this heterojunction device to process a 64×64 color image containing Gaussian noise. DETAILED DESCRIPTION

[0040] The following are specific embodiments of the present invention, which further illustrate the present invention. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application.

[0041] like Figure 1As shown in FIG. 1 , a bipolar two-dimensional semiconductor molybdenum telluride (MoTe2) is prepared on a P-type heavily doped silicon wafer with silicon oxide deposited on it, and a heterojunction with an N-type two-dimensional semiconductor molybdenum sulfide (MoS2) on it is used as a color image sensor. The color image sensor can simultaneously realize the acquisition and processing of image data with convolution calculation through the combination of back gate voltage and source-drain voltage. Figure 2 As shown, the application of the color image sensor in image convolution processing is described in detail below.

[0042] Furthermore, the dynamic range of the color image sensor is pre-calibrated through experiments, that is, the upper and lower limits of the light signal intensity that can be detected, and the upper limit is recorded as P max , the lower limit is denoted as P min The desired grayscale value range of the image is denoted as G. The mapping relationship between the light signal intensity and the grayscale value range is established according to formula (1):

[0043] P=qG (1)

[0044] Where P is P max With P min The difference is , q is the proportional coefficient.

[0045] Furthermore, a total of c channel types are determined according to the color that needs to be paid attention to, which are recorded as C1, C2, ..., C i ,…,C c , i=1,2,…,c, where C i Represents the wavelength band corresponding to channel i. For example, if you need to focus on blue, you can select C in the range of 450nm to 520nm. i .

[0046] Furthermore, when using C1 to C c The transfer curve of the device is tested under the condition that the device is illuminated by a light signal in the wavelength range, and the corresponding gate voltage V is extracted. g1 ,V g2 ,…,V gi ,…,V gc , i=1,2,…,c, where V gi This means that at this gate voltage, the device is i The responsiveness is the highest.

[0047] Furthermore, for the data of c channels, the corresponding c convolution kernels K are determined i, i=1,2,…,c, the way to determine these convolution kernels is related to the purpose of image convolution processing. For example, if you want to extract the edge of the target, use the Sobel convolution kernel; if you want to sharpen the image, use the Laplace convolution kernel; if you want to Gaussian filter the image, use the Gaussian convolution kernel. The Gaussian convolution kernel is a two-dimensional square matrix with a size of m×m, where m is an odd number, usually 3. Then, the standard deviation σ of the convolution kernel is determined according to the expected processing amplitude, for example, 2. After determining m and σ, the weight W of each position in the convolution kernel under each band can be determined. xy i , x=1,2,…,m, y=1,2,…,m, i=1,2,…,c.

[0048] Furthermore, for channel C i , apply gate voltage V to the device gi , using C i The optical signal of the wavelength band illuminates the device and scans the source-drain voltage V ds To test the output curve of the device and extract the photocurrent I ph Output curve, V ds The selection basis is to enable the device to exhibit complete rectification characteristics, and its range is set to (V ds，min ,V ds，max ), in P max With P min Internal adjustment C i The intensity of the optical signal in the band, that is, the intensity is P1, P2, ..., P j ,…,P n , j=1,2,…,n i The band optical signal test obtains n groups of photocurrents I pj For the output curve, n needs to select a suitable value, which should be at least greater than 4. The larger the value, the more conducive it is to improving the accuracy of the subsequent fitting process.

[0049] Furthermore, for different strengths of C i n groups of photocurrents I obtained from the band optical signal pj The output curve is fitted using formula (2):

[0050] I ph =α(kP) β (2)

[0051] Where P is the light signal intensity, α is the proportionality coefficient, and k is the ds function, which is related to V ds is positively correlated, and its range of variation represents V ds The ability to adjust the output signal when the light intensity (optical power density) is fixed, and β reflects the linear relationship between the photocurrent and the input light intensity. iIn the case of a light signal, α and β are fixed values, that is, they are independent of the light intensity and only related to the wavelength. Repeat this step to obtain each C i The corresponding α and β values.

[0052] Furthermore, V ds，max The corresponding k value is the largest, which is k max , while V ds，min The corresponding k value is the smallest, which is k min , for all values ​​of k, divide by k max To complete the normalization, the normalized k value is regarded as the weight W, and the fitting of each C is obtained according to formula (3): i W and V under ds Relationship:

[0053]

[0054] Among them, a and b reflect the change of W with V ds The speed of change, at a certain C i Under the optical signal, a and b are fixed values. Repeat this step to obtain each C i The corresponding a and b values.

[0055] Furthermore, the same device is divided into two groups according to the convolution kernel K. i The pixels are arranged in a way to form a photosensitive array. Each pixel corresponds to a device. All devices are connected in parallel. i The weight W of each pixel in xy i Normalize and then substitute into formula (3) to get the source-drain voltage V that needs to be applied to the device at the corresponding position of the pixel ds，xy , so that each C i Next, V of size m×m ds Matrix, denoted by V ds i .

[0056] Furthermore, in practical applications, in order to perform convolution processing on the initial image to obtain a feature map, first, for the selected C1~C c Each C under the wavelength range i , apply V to the gate of each device in the photosensitive array gi , and the source-drain voltage is determined based on the K i The corresponding matrix V ds i To apply, read the sum of the currents of m×m parallel devices, and then maintain V gi The convolution kernel remains unchanged, and the entire initial image is traversed. The sum of the currents is read multiple times. The sum of the read currents is substituted into formula (2) to obtain the light intensity P, and then C is obtained according to formula (1). iThe gray value distribution diagram at this time is the gray value distribution diagram at C i The feature map processed by convolution in this band, combined with C1~C c All feature maps within the wavelength range can obtain the final multi-band color image after convolution processing.

[0057] Example 1

[0058] In this embodiment, the structure of the heterojunction device based on the above-mentioned bipolar and N-type two-dimensional semiconductors is as follows: Figure 1 As shown, the substrate is a P-type heavily doped silicon wafer (resistivity 0.001-0.005Ω·cm) with a 285nm thick silicon oxide deposited on it. The bipolar MoTe2 and N-type MoS2 films come from bulk materials that have been mechanically stripped, and are stacked on silicon oxide in turn using dry transfer technology. The window area where MoTe2 contacts the metal electrode is cleaned with an oxygen plasma of 60W and 50sccm for 1min before deposition. After that, the source and drain areas are determined by graphics, and a 40nm thick gold film is deposited by thermal evaporation as the contact electrode. P-type heavily doped silicon is used as the back gate. The device can simultaneously realize the perception and convolution processing of multi-band optical signals at the device level.

[0059] Example 2

[0060] In this embodiment, based on the heterojunction device constructed by the bipolar and N-type two-dimensional semiconductors, the heterojunction is irradiated with 450-700nm lasers, and the photocurrent is tested under the conditions of source-drain voltage of 2V and gate scanning range of -50V to 50V. The responsivity is then calculated and normalized according to the light intensity of different bands used. Figure 3 The normalized responsivity distribution is shown. As the gate voltage increases, the wavelength corresponding to the peak responsivity gradually transitions from 450nm to around 750nm, realizing the reconstruction of the spectral response function of the device under different gate voltages.

[0061] Example 3

[0062] In this embodiment, based on the results of the above-mentioned embodiment 2, the gate voltage is selected to be -50V and the source-drain voltage is selected to be 2V to mainly sense the laser with a wavelength band of 450nm, that is, the blue channel; the gate voltage is selected to be 0V and the source-drain voltage is selected to be 2V to mainly sense the laser with a wavelength band of 700nm, that is, the red channel. Figure 4 The perception of the example blue letter "I" and the red letter "C" are demonstrated respectively. By comparing the intensity distribution of the initial image and the imaging effect under two gate voltages, it can be seen that when using different gate voltages, the light signal in the corresponding main response band can be made more obvious, while the light signal in the non-main response band can be made weaker.

[0063] Example 4

[0064] In this embodiment, a heterojunction device based on the above-mentioned bipolar and N-type two-dimensional semiconductors is constructed to demonstrate the use of V ds The matrix realizes the convolution calculation process. Figure 5 As shown in , a Gaussian convolution kernel with a size of 3×3 and a standard deviation of 2 is selected, and the weight corresponding to each pixel is calculated after normalization. Figure 2 The process shown in the figure obtains the coefficients in formula (3). Substituting the weights into formula (3) obtains V corresponding to the blue and red channels respectively. ds matrix.

[0065] Example 5

[0066] In this embodiment, the V corresponding to the blue and red channels obtained based on Example 4 ds Matrix, the source-drain voltage of the heterojunction device at the corresponding position is calculated according to V ds Matrix is ​​applied. Figure 6 As shown, taking a 64×64 color image with Gaussian noise as an example, the gate voltage is -50V, and the V of the blue channel corresponding to the Gaussian convolution kernel is ds Matrix to collect blue channel image; use gate voltage of 0V and Gaussian convolution kernel corresponding to the red channel V ds The color initial image is divided into blue and red channels and Gaussian convolution is performed to obtain a significantly reduced output image noise.

[0067] The specific embodiments described herein are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the appended claims.

Claims

1. A color image sensor based on a two-dimensional semiconductor heterostructure, characterized in that: It includes a bipolar two-dimensional semiconductor layer and an N-type two-dimensional semiconductor layer deposited on a substrate, which form a horizontal heterojunction. The color image sensor can simultaneously realize the collection and processing of image data with convolution calculation through a combination of gate voltage and source-drain voltage.

2. The color image sensor based on a two-dimensional semiconductor heterostructure according to claim 1, characterized in that: The substrate is a P-type heavily doped silicon wafer with deposited silicon oxide; the material of the bipolar two-dimensional semiconductor layer is molybdenum telluride; and the material of the N-type two-dimensional semiconductor layer is molybdenum sulfide.

3. The color image sensor based on a two-dimensional semiconductor heterostructure according to claim 1, characterized in that: The energy band arrangement of the heterojunction is controlled by adjusting the gate voltage, thereby realizing different spectral perception functions and collecting multi-channel image information without the need for color filters.

4. The color image sensor based on a two-dimensional semiconductor heterostructure according to claim 1, characterized in that: The photocurrent response intensity of the heterojunction is controlled by adjusting the source-drain voltage, thereby giving different channel weights to light signals of different wavelengths, thereby realizing the convolution calculation process of the image.

5. An application of a color image sensor based on a two-dimensional semiconductor heterostructure according to any one of claims 1 to 4 in image convolution processing, characterized in that: include: Determine the range of optical signal intensity and channel type, and determine the mapping relationship between optical signal intensity and grayscale value; Determine the channel type according to the color that needs to be paid attention to, and determine the gate voltage of the device according to the channel type; Determine the convolution kernel corresponding to each channel according to the purpose of image convolution processing, calculate the weight distribution of the convolution kernel in each band and normalize it; Establish the mapping relationship between device photocurrent and light intensity under the specified channel, as well as the mapping relationship between weight and source-drain voltage; Arrange several identical devices according to the size and shape of the convolution kernel and connect them in parallel to form a photosensitive array; Determine the source-drain voltage matrix of the photosensitive array according to the convolution kernel weight distribution and mapping relationship; When collecting images, a predetermined matrix of gate voltage and source-drain voltage under different channels is applied to the photosensitive array to read the photocurrent and; The photocurrent and the grayscale value are converted to obtain the images after convolution processing under different channels, and the color image is obtained by fusion.

6. The use according to claim 5, characterized in that: The mapping relationship between the light signal intensity and the grayscale value is specifically determined as follows: the dynamic range of the color image sensor is pre-calibrated to obtain the upper limit P of the light signal intensity that can be detected. max and the lower limit P min ; The expected grayscale value range of the image is recorded as G. The mapping relationship between the light signal intensity and the grayscale value is established according to the following formula: P = qG, where P is P max With P min The difference is , q is the proportional coefficient.

7. The use according to claim 6, characterized in that: The gate voltage is determined specifically as follows: According to the color that needs to be paid attention to, a total of c channel types are determined, denoted as C1, C2, ..., C i ,…,C c , i=1,2,…,c, where C i Represents the band corresponding to channel i; When using C1~C c The transfer curve of the device is tested under the condition that the device is illuminated by a light signal in the wavelength range, and the corresponding gate voltage V is extracted. g1 ,V g2 ,…,V gi ,…,V gc , i=1,2,…,c, where V gi This means that at this gate voltage, the device is i The responsiveness is the highest.

8. The use according to claim 7, characterized in that: For channel C i , apply gate voltage V to the device gi , using C i The optical signal of the wavelength band illuminates the device and scans the source-drain voltage V ds To test the output curve of the device and extract the photocurrent I ph Output curve, V ds The selection basis is to enable the device to exhibit complete rectification characteristics, and its range is set to (V ds，min ,V ds，max ), in P max With P min Internal adjustment C i The intensity of the light signal in the band is used to obtain multiple groups of photocurrents I ph The output curve is fitted using the following formula: The ph =α(kP) β Where P is the light signal intensity, α is the proportionality coefficient, and k is the ds function, which is related to V ds is positively correlated, and its range of variation represents V ds The ability to adjust the output signal when the light intensity is fixed, while β reflects the linear relationship between the photocurrent and the input light intensity. i In the case of band optical signals, α and β are fixed values. Repeat this step to obtain the value of each C i The corresponding α and β values; Among them, V ds，max The corresponding maximum k value is recorded as k max , while V ds，min The corresponding minimum k value is recorded as k min , for all values ​​of k, divide by k max To complete the normalization, the normalized k value is regarded as the weight W, and the following formula is used to fit each C i W and V under ds Relationship: Among them, a and b reflect the change of W with V ds The speed of change of C i In the case of band optical signals, a and b are fixed values. Repeat this step to obtain each C i The corresponding a and b values.

9. The use according to claim 8, characterized in that: The same devices are arranged according to the pixel arrangement of the convolution kernel to form a photosensitive array. Each pixel corresponds to a device, and all devices are connected in parallel. i The corresponding convolution kernel K i The weight of each pixel in is normalized and then substituted into the fitted channel C i Corresponding W and V ds , the source-drain voltage required to be applied to the device at the corresponding position of the pixel is obtained, thereby obtaining each C i The source-drain voltage V ds Matrix, denoted by V ds i .

10. The use according to claim 9, characterized in that: In order to convolve the initial image to obtain the feature map, first, for the selected C1~C c Each C under the wavelength range i , apply V to the gate of each device in the photosensitive array gi , the source-drain voltage is based on the matrix V ds i to apply, read the sum of the currents of all parallel devices, and then maintain V gi The convolution kernel remains unchanged, traverses the entire initial image, reads the current sum multiple times, and converts the read current sum into C i The gray value distribution diagram obtained is the gray value distribution diagram under C i The convolutional feature map under the band, combined with C1~C c All feature maps under the wavelength range are finally convolved to obtain a multi-band color image.

Images