CN113947536A - Self-adaptive enhancement method for heterologous night vision halation image - Google Patents
Self-adaptive enhancement method for heterologous night vision halation image Download PDFInfo
- Publication number
- CN113947536A CN113947536A CN202111054226.6A CN202111054226A CN113947536A CN 113947536 A CN113947536 A CN 113947536A CN 202111054226 A CN202111054226 A CN 202111054226A CN 113947536 A CN113947536 A CN 113947536A
- Authority
- CN
- China
- Prior art keywords
- image
- halation
- night vision
- adaptive
- gray value
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000004297 night vision Effects 0.000 title claims abstract description 78
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000002834 transmittance Methods 0.000 claims abstract description 35
- 230000002708 enhancing effect Effects 0.000 claims abstract description 20
- 230000003044 adaptive effect Effects 0.000 claims description 50
- 238000011156 evaluation Methods 0.000 claims description 12
- 125000001475 halogen functional group Chemical group 0.000 claims description 10
- 230000000875 corresponding effect Effects 0.000 claims description 8
- 230000008569 process Effects 0.000 claims description 5
- 206010019099 Halo vision Diseases 0.000 claims description 4
- 238000001914 filtration Methods 0.000 claims description 4
- 241000764238 Isis Species 0.000 claims description 3
- 230000002596 correlated effect Effects 0.000 claims description 3
- 238000011160 research Methods 0.000 claims description 3
- 230000011218 segmentation Effects 0.000 claims description 3
- 230000004927 fusion Effects 0.000 abstract description 53
- 238000003384 imaging method Methods 0.000 abstract description 7
- 230000000694 effects Effects 0.000 description 13
- 230000006872 improvement Effects 0.000 description 12
- 238000005286 illumination Methods 0.000 description 8
- 230000008030 elimination Effects 0.000 description 6
- 238000003379 elimination reaction Methods 0.000 description 6
- 238000002474 experimental method Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 230000000007 visual effect Effects 0.000 description 3
- 238000004088 simulation Methods 0.000 description 2
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 230000005888 antibody-dependent cellular phagocytosis Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000007499 fusion processing Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 210000001525 retina Anatomy 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
- G06T5/92—Dynamic range modification of images or parts thereof based on global image properties
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a self-adaptive enhancement method of a heterogeneous night vision halation image, which solves the problem that the existing image enhancement algorithm in the prior art is not suitable for enhancing the night vision halation image, can effectively improve the dark detailed information of a visible image and the definition of an infrared image in a halation scene, simultaneously avoids the excessive enhancement of an image halation area, and improves the imaging quality of an anti-halation fusion image. The invention comprises the following steps: step 1, determining a halation critical gray value G of a night vision visible light image; step 2, negating the night vision halation image, and estimating the initial transmittance of the negated image; step 3, constructing a self-adaptive transmittance function T according to the initial transmittance of the image and the halation critical gray value G; and 4, carrying out self-adaptive enhancement on the images with different halation degrees according to the self-adaptive transmittance function T.
Description
The technical field is as follows:
the invention belongs to the technical field of night vision anti-halation of automobiles, and relates to a self-adaptive enhancement method of a heterologous night vision halation image, which is used for improving the definition and contrast of a visible light image and an infrared image in a halation scene and improving the imaging quality of an anti-halation fusion image.
Background art:
when a driver drives at night, the driver is irradiated by a strong light source of an opposite vehicle to generate a serious halation phenomenon, so that the driver cannot see the road condition ahead clearly in a short time, and traffic accidents are easily caused. The heterogeneous night vision image fusion anti-halation technology utilizes the advantages that an infrared image is not halation, and visible light images are rich in color detail information, and infrared and visible light images are fused for anti-halation, so that the obtained fused image halation is eliminated more thoroughly, the image visual effect is better, and the safety of driving at night can be effectively improved. However, due to the special night vision vignetting scene, in the preprocessing stage of image fusion, the existing image enhancement algorithm amplifies noise and vignetting area in the image while enhancing the detail information of the dark part of the image when processing the night vision vignetting image, so that the effective information improvement effect is not obvious, the imaging quality of the fused image is influenced, and the method is not suitable for enhancing the night vision vignetting image.
The existing enhancement algorithm based on Retinex theory can effectively improve the definition of an image, but distortion is easily generated in the enhancement of a low-illumination image with insufficient illumination, noise in the image is amplified in the enhancement process, and the obtained anti-halation image outline and color detail improvement effect is not obvious; the local Histogram Equalization (HE) enhancement algorithm has a good effect on enhancing images with over-bright or over-dark foreground and background, but is easy to enlarge the halation area in the images when used for enhancing low-illumination images with halation; the homomorphic filtering enhancement algorithm is suitable for enhancing an image with uniform environmental illumination, and is not good in effect of enhancing the image at night and is not suitable for enhancing a low-illumination image with halation; the dark channel prior enhancement algorithm (DCP) can enhance dark information in an image, has obvious enhancement effect, is suitable for enhancing a low-illumination image, and can enlarge a halation area in the image when being used for enhancing the low-illumination image with halation. In summary, several types of image enhancement algorithms are not suitable for enhancing the night vision halation image.
The invention content is as follows:
the invention aims to provide a self-adaptive enhancement method of a heterogeneous night vision halation image, which solves the problem that the existing image enhancement algorithm in the prior art is not suitable for enhancing the night vision halation image, can effectively improve the dark detailed information of a visible image and the definition of an infrared image in a halation scene, and improves the imaging quality of an anti-halation fusion image; the method avoids the excessive enhancement of the image halo area while realizing the enhancement of the detail information of the image dark place, can realize the self-adaptive enhancement of the images with different halo degrees, and is used for the image enhancement under the night vision halo scene.
In order to achieve the purpose, the invention adopts the technical scheme that:
a self-adaptive enhancement method of a heterogeneous night vision halation image is characterized by comprising the following steps: the method comprises the following steps:
step 1, determining halation critical gray value G of night vision visible light imagec；
Step 2, negating the night vision halation image, and estimating the initial transmissivity t of the negated image;
step 3, according to the initial transmissivity t and the halation critical gray value G of the imagecConstructing an adaptive transmittance function T;
when the vignetting degree of the night vision vignetting image is larger, the image transmissivity is properly reduced; according to the functional relation between a large number of night vision halation images with the halation degrees from small to large and the corresponding optimal image transmissivity, an adaptive transmissivity function T is constructed by means of least square nonlinear fitting, wherein the adaptive transmissivity function T is as follows:
in the formula, x is a pixel gray value of the night vision halation image;
in the process of enhancing the night vision halation image, the images with different halation degrees adopt corresponding transmissivity to adaptively enhance the night vision halation image; when halation imageThe gray value of the pixel is less than the halo critical gray value GcThen, the initial transmittance t is used; when the pixel gray value x of the halation image is larger than or equal to the halation critical gray value GcThe transmittance is adjusted using the constructed adaptive transmittance function.
And 4, carrying out self-adaptive enhancement on the images with different halation degrees according to the self-adaptive transmittance function T.
In step 1, determining a halation critical gray value G of a night vision visible light imagecThe method comprises the following steps:
step 1.1, calculating the inter-class squared difference theta according to the following formula, and obtaining an initial critical gray value k when theta is the maximum value:
in the formula, mGExpressed as the average gray value of the whole image
P1(k) For the probability of the gray level in the image, the expression is
Is the average gray value of pixels with gray levels from 0 to k
Step 1.2, calculating a halation critical gray value G according to the following formulac：
Gc＝k+s (2)
In the formula, s is a superposition coefficient.
The function expression of the superposition coefficient s in the formula (2) is realized by the following steps:
(1) collecting points: determining an initial critical gray value k according to a maximum inter-class variance method by taking halo images with different halo degrees as research objects, and obtaining a superposition coefficient s by performing optimal threshold segmentation on the night vision halo image according to the superposition coefficient of the initial critical gray value; collectingCorresponding point set (k) of a large number of vignetting images of different types of road vehicles from far and near and with vignetting decreasing from small to largei，si)；
(2) Linear fitting: according to point set (k)i，si) The superposition coefficient s and the initial critical gray value k are inversely correlated, and least square normal fitting is carried out by adopting the following formula:
s＝a*k+b (3)
(3) determining the value of the coefficient: when the evaluation function L (a, b) is set to the minimum value, the values of a and b in the equation s are set to-1.11 and 181.14, respectively.
In step 2, the initial transmittance t of the inverted image is estimated according to the following formula:
in the formula, omega is a defogging coefficient; i (x) is a vignetting inversion image; a is the atmospheric light intensity at infinity; avref denotes mean filtering.
In step 4, the transmittance of the halation image is determined according to the adaptive transmittance function T, and the enhanced low-illuminance image j (x) is obtained by combining the following formula:
the low-intensity image j (x) is re-inverted to obtain an enhanced vignetting image.
Compared with the prior art, the invention has the advantages and effects that:
1. the self-adaptive enhancement algorithm of the heterogeneous night vision halation image designed by the invention improves the detail information of the dark part of the visible image and the definition of the infrared image in a halation scene, simultaneously avoids the excessive enhancement of the halation area of the visible image, and improves the imaging quality of the anti-halation fusion image.
2. The image halation critical gray value designed by the invention can be determined according to the halation degree of the halation image by adopting an improved maximum inter-class variance method, and the image is divided into a halation area and a non-halation area.
3. The image adaptive enhancement algorithm designed by the invention can adopt different transmittances according to different halation degrees of the night vision halation image, construct an adaptive transmittance function according to the functional relation between the initial transmittance and the halation critical gray value, and automatically adjust along with the halation degree of the image, thereby realizing the adaptive enhancement of the images with different halation degrees.
Description of the drawings:
FIG. 1(a) is a visible light original drawing of a faint scene in a small area of a trunk road in a city;
FIG. 1(b) is an infrared original drawing of a small-area halation scene of a trunk road in a city;
fig. 2(a) is a visible light original image of a suburban road small-area halation scene;
FIG. 2(b) is an infrared original image of a suburban road small-area halation scene;
fig. 3(a) is a visible light original image of a large-area halo scene of a suburb trunk road;
FIG. 3(b) is an infrared original image of a large-area halo scene of a suburb trunk road;
FIG. 4(a) is a visible light original image of a large-area halation scene of a suburb road;
FIG. 4(b) is an infrared original image of a large-area halation scene of a suburb road;
FIG. 5(a) is a visible light image of a small-area vignetting scene of an urban trunk road enhanced by an adaptive enhancement algorithm;
FIG. 5(b) is an infrared image of a small-area halation scene of a trunk road in a city enhanced by an adaptive enhancement algorithm;
FIG. 6(a) is a visible light image of a suburban road small area vignetting scene enhanced by an adaptive enhancement algorithm;
FIG. 6(b) is an infrared image of a suburban road small area vignetting scene enhanced by an adaptive enhancement algorithm;
FIG. 7(a) is a visible light image of a large-area vignetting scene of a suburban trunk road enhanced by an adaptive enhancement algorithm;
FIG. 7(b) is an infrared image of a large-area halation scene of a suburban trunk road enhanced by an adaptive enhancement algorithm;
FIG. 8(a) is a visible light image of a suburban road large area vignetting scene enhanced by an adaptive enhancement algorithm;
FIG. 8(b) is an infrared image of a suburban road large area vignetting scene enhanced by an adaptive enhancement algorithm;
FIG. 9 is an anti-halation image fused after enhancement of the self-adaptive enhancement algorithm for halation images in small areas of urban trunk roads;
FIG. 10 is a fused anti-halation image after the suburban road small area halation image adaptive enhancement algorithm is enhanced;
FIG. 11 is an anti-halation image fused after the adaptive enhancement algorithm for the halation image of the large area of the suburban trunk road is enhanced;
FIG. 12 is a fused anti-halation image after the suburb road large area halation image adaptive enhancement algorithm is enhanced;
the specific implementation mode is as follows:
in order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Aiming at the problem that the existing image enhancement algorithm is not suitable for enhancing the night vision vignetting image, the invention provides a self-adaptive enhancement algorithm of a different-source night vision vignetting image, and the imaging quality of an anti-vignetting fusion image is improved.
The invention is described in detail below with reference to the drawings and a complete process.
Aiming at the fact that the existing image enhancement algorithm is not suitable for enhancing the night vision halation image, the invention designs a self-adaptive enhancement algorithm of a different-source night vision halation image. Since the transmission of the vignetting image determines the degree of vignetting of the vignetting image, the degree of vignetting of the night vision vignetting image can be reflected by the transmission of the image. In order to enhance the night vision halation image more effectively, the method constructs an adaptive transmittance function by researching the functional relation between the halation degree of the night vision halation image and the transmittance of an enhancement algorithm, automatically adjusts the transmittance of the image according to the halation degree of the night vision halation image, and achieves the adaptive enhancement of images with different halation degrees. The method improves the detail information of the dark part of the visible light image and the definition of the infrared image in the halation scene, and improves the imaging quality of the anti-halation fusion image. The night vision halation image can be well enhanced, the image dark position is enhanced, meanwhile, transition enhancement of an image halation area is avoided, and self-adaptive enhancement of images with different halation degrees is achieved. The invention is suitable for the technical field of automobile night vision anti-halation.
The invention provides a self-adaptive enhancement method of a heterologous night vision halation image, which comprises the following steps:
step 1, determining halation critical gray value G of night vision visible light imagecThe method comprises the following processing steps:
step 1.1, calculating the inter-class squared difference theta according to the following formula, and obtaining an initial critical gray value k when theta is the maximum value:
in the formula, mGExpressed as the average gray value of the whole image
P1(k) For the probability of the gray level in the image, the expression is
Is the average gray value of pixels with gray levels from 0 to k
Step 1.2, calculating a halation critical gray value G according to the following formulac：
Gc＝k+s#(2)
In the formula: and s is a superposition coefficient.
The function expression of the superposition coefficient s in the formula (2) is specifically realized by the following steps:
1. collecting points: and determining an initial critical gray value k according to a maximum inter-class variance method by taking the vignetting images with different vignetting degrees as research objects, and obtaining a superposition coefficient s by performing optimal threshold segmentation on the night vision vignetting images according to the superposition coefficient of the initial critical gray value. Corresponding point sets (k) of a large number of halation images of different types of road vehicles from far and near and with the halation becoming smaller from small to large are collectedi，si)；
2. Linear fitting: according to point set (k)i，si) The superposition coefficient s and the initial critical gray value k are inversely correlated, and least square normal fitting is carried out by adopting the following formula:
s＝a*k+b#(3)
3. determining the value of the coefficient: when the evaluation function L (a, b) is set to the minimum value, the values of a and b in the equation s are set to-1.11 and 181.14, respectively.
Step 2, inverting the night vision halation image, and estimating the initial transmittance y of the inverted image, wherein the method comprises the following processing steps:
step 2.1, reversing the night vision halation image:
J(x)＝255-I(x)#(5)
step 2.2, estimating the initial transmittance t of the inverted image according to the following formula:
in the formula: omega is defogging coefficient; i (x) is a vignetting inversion image; a is the atmospheric light intensity at infinity; avref denotes mean filtering;
step 3, according to the initial transmissivity t and the halation critical gray value G of the imagecAn adaptive transmittance function T is constructed.
The transmission of the vignetting image determines the degree of vignetting of the vignetting image. In order to obtain a halation image with a better enhancement effect, the image transmittance is appropriately reduced when the degree of halation of the night vision halation image is large. According to the relation between a large number of night vision halation images with the halation degrees from small to large and the corresponding optimal image transmissivity, an adaptive transmissivity function T is constructed through least square nonlinear fitting and is as follows:
in the formula, x is the pixel gray value of the night vision halation image.
In the process of enhancing the night vision halation image, the images with different halation degrees adopt corresponding transmissivity to perform self-adaptive enhancement on the night vision halation image. When the gray value of the pixel of the halation image is less than the critical gray value G of the halationcThen, the initial transmittance t is used; when the pixel gray value x of the halation image is larger than or equal to the halation critical gray value GcThe transmittance is adjusted using the constructed adaptive transmittance function.
And 4, carrying out self-adaptive enhancement on the images with different halation degrees according to the self-adaptive transmittance function T.
Determining the transmissivity of the halation image according to the adaptive transmissivity function T, and combining the formula to obtain an enhanced low-illumination image J (x):
the low-intensity image j (x) is re-inverted to obtain an enhanced vignetting image.
Example (b):
an example of a specific simulation is given below.
Simulation conditions of this embodiment: windows10 operating system, MATLAB software.
The main contents are as follows: the method comprises the steps of determining a halation critical gray value of a halation image by adopting an improved maximum inter-class variance method, negating the night vision halation image, estimating initial transmissivity of the negated image, constructing a self-adaptive transmissivity function by combining the halation critical gray value of the halation image, the negated image and the initial transmissivity, automatically adjusting the image transmissivity according to the halation degree of the image, and realizing self-adaptive enhancement of the images with different halation degrees. The method comprises the following specific steps:
firstly, calculating a halation critical gray value:
1. reading visible light and infrared original images in a halation scene by using an imread function;
2. converting the visible light color image into a gray image by using an rgb2gray function;
3. calculating the halation critical gray value G of the halation image according to the formulas (1) and (2)c；
Secondly, adaptively enhancing visible light and infrared images in a night vision halation scene:
1. negating the vignetting image according to formula (5), and calculating the initial transmittance t according to formula (6);
2. adaptively enhancing visible light and infrared reflection images of 4 groups of halation scene images according to a transmissivity function T constructed by a formula (7);
3. and combining the formula (8) and performing further inversion to obtain 4 groups of enhanced vignetting scene images.
Thirdly, enhancing and analyzing the night vision halation scene image:
in order to verify the effectiveness of the adaptive enhancement algorithm of the heterologous night vision halation image in fusing anti-halation and the universality of the heterologous night vision halation image in different halation scenes, the adaptive enhancement algorithm provided by the invention has the advantages that the adaptive enhancement algorithm can be used for improving the compatibility of the heterologous night vision halation image in the fusion anti-halation scene, image enhancement experiments are carried out on 4 groups of visible light and infrared images under different night vision halation scenes, namely visible light and infrared original images of a small-area halation scene of a trunk road in a city (see fig. 1(a) and 1(b)), visible light and infrared original images of a small-area halation scene of a suburban road (see fig. 2(a) and 2(b)), visible light and infrared original images of a large-area halation scene of a suburban trunk road (see fig. 3(a) and 3(b)), and visible light and infrared original images of a large-area halation scene of a suburban road (see fig. 4(a) and 4 (b)). The comparison experiment is carried out on the visible light images and the infrared images under 4 groups of different halation scenes by respectively adopting Histogram Equalization (HE), multi-scale retina (MSR), dark channel prior enhancement algorithm (DCP) and the Adaptive Enhancement Algorithm (AEA) of the invention. And performing objective analysis on the enhanced visible light and infrared images by using indexes such as average absolute error (MAE), Average Gradient (AG), Information Entropy (IE), peak signal-to-noise ratio (PSNR), Structural Similarity (SSIM) and the like.
The visible light image and the infrared image of the urban trunk road small-area halation scene enhanced by the Adaptive Enhancement Algorithm (AEA) of the invention are shown in fig. 5(a) and 5 (b).
Night vision image enhancement experiment results of urban main road small area halation are shown in table 1.
TABLE 1 Objective index data of enhanced image in small area of lightsickness scene on urban arterial roads
As can be seen from the objective index data of the enhanced image in the urban trunk road halation scene in Table 1, the MAE index of the visible light image enhanced by the algorithm is reduced by 86.32% and 50.04% compared with HE and MSR indexes, and is increased by 11.64% compared with a DCP enhanced image, which indicates that the brightness of the image enhanced by the HE and MSR algorithms is increased too high, the brightness of the image enhanced by the DCP algorithm is increased less, and the brightness of the image enhanced by the algorithm is increased moderately; as can be seen from the peak signal-to-noise ratio and the structural similarity index data, the PSNR and SSIM indexes of the enhanced images processed by the algorithm are obviously improved, and the enhanced infrared images are improved by 51.37 percent and 34.98 percent compared with the HE and MSR enhanced images, which shows that the image quality enhanced by the algorithm is better than the improvement effect of the original image.
Visible light images and infrared images of a small-area halation scene of a suburban road enhanced by the Adaptive Enhancement Algorithm (AEA) of the present invention are shown in fig. 6(a) and 6 (b).
The result of the small-area halation night vision image enhancement experiment of suburban roads is shown in table 2.
TABLE 2 Objective index data of enhanced images for small area lightsickness scenes on suburb roads
As can be seen from the objective index data of the enhanced image of the small-area halation scene on the suburb road in table 2, the IE index of the visible light image enhanced by the algorithm is improved by 26.87% and 44.97% compared with the HE and MSR enhanced images, which indicates that the image enhanced by the algorithm of the invention contains more abundant information; the PSNR index of the infrared image enhanced by the algorithm is improved by 39.12 percent and 23.57 percent compared with MSR and DCP enhanced images, and the peak signal-to-noise ratio is obviously improved, so that the enhanced image has smaller noise compared with the original image, and the image quality is improved.
The visible light image and the infrared image of the large-area halo scene of the suburban trunk road are enhanced by the Adaptive Enhancement Algorithm (AEA) of the invention as shown in fig. 7(a) and 7 (b).
The night vision infrared and visible light image enhancement result of the large-area halation of the suburb trunk road is compared with that shown in the table 3.
TABLE 3 Objective index data of enhanced image in large-area halation scene on suburb arterial road
As can be seen from the large-area halation scene enhanced image index data in table 3, the enhanced image value of the halation in the area with smaller enhanced visible light and infrared image MAE index values is high, which indicates that the brightness of the enhanced image is affected by the halation area of the image, and the image quality is reduced when the image brightness is too high or too low, but the MAE index of the algorithm enhanced image is obviously reduced compared with HE and MSR and is improved compared with DCP, and the proper improvement of the image brightness is helpful for improving the visibility of the information in the dark area of the image; indexes such as AG, IE, PSNR, SSIM and the like of the image enhanced by the algorithm are higher than those of the HE, MSR and DCP enhanced algorithms, and the image enhanced by the algorithm is clearer, contains more information and has better quality after enhancement.
The visible light image and the infrared image of the large-area halation scene of the suburban road enhanced by the Adaptive Enhancement Algorithm (AEA) of the present invention are shown in fig. 8(a) and 8 (b).
The night vision infrared and visible light image enhancement results of the large-area halation of the suburb road are compared in a table 4.
TABLE 4 Objective index data of enhanced image of large area halation scene on suburb community road
As can be seen from the large-area halation scene enhanced image index data in table 4, the value of the halation enhanced image in the area with smaller MAE index values of the enhanced visible light and infrared images is high, which indicates that the brightness of the enhanced image is affected by the halation area of the image, and the image quality is reduced when the image brightness is too high or too low, but the MAE index of the algorithm enhanced image is obviously reduced compared with HE and MSR, and is improved compared with DCP, and the proper improvement of the image brightness is beneficial to improving the visibility of the information in the dark area of the image; indexes such as AG, IE, PSNR, SSIM and the like of the image enhanced by the algorithm are higher than those of the HE, MSR and DCP enhanced algorithms, and the image enhanced by the algorithm is clearer, contains more information and has better quality after enhancement.
The experimental results and analysis of the visible light and infrared image enhancement under different halation scenes show that the visible light and infrared image enhanced by the adaptive enhancement algorithm are moderately enhanced in brightness, the halation area is not amplified while the brightness of the dark part of the visible light image is enhanced, the definition of the enhanced infrared image is obviously improved, noise is not generated, and buildings and roadside trees in the image are clearly visible. The objective index data of the enhanced image shows that the enhancement algorithm of the invention has moderate improvement on the brightness of the image, avoids over-enhancement of the halation part, and improves the integral definition and visibility of the image.
Four, night vision image fusion anti-halation result and analysis
In order to fully illustrate the improvement effect of the self-adaptive enhancement algorithm of the heterogeneous night vision halation image in anti-halation image fusion, an anti-halation fusion experiment is carried out on the visible light and the infrared image which are processed by the enhancement algorithm in 4 kinds of halation scenes, the same improved IHS-Curvelet fusion algorithm is adopted in the image fusion process, and the enhancement algorithm is only partially different.
The anti-halation image fused by the improved IHS-Curvelet fusion algorithm is shown in FIG. 9, wherein the visible light image and the infrared image of the halation scene in the small region of the urban trunk road are enhanced by the Adaptive Enhancement Algorithm (AEA).
The evaluation index data pairs of the urban trunk road small-area halation fusion image are shown in table 5.
TABLE 5 urban trunk road small region halation fusion image evaluation index
As can be seen from the index data in Table 5, the fused image vignetting elimination degree D index after the enhancement of the HE algorithm and the DCP algorithm is smaller, which indicates that the fused image vignetting elimination effect after the enhancement of the HE algorithm and the DCP algorithm is poorer, and the fused image vignetting elimination degree index of the MSR enhancement and the self-adaptive enhancement of the invention is higher, wherein the vignetting elimination degree value of the enhanced fused image of the invention is maximum, which indicates that the vignetting elimination of the fused image is more thorough. Compared with HE fusion, MSR fusion and DCP fusion, the fusion image obtained by the enhancement algorithm of the invention has improved indexes, wherein the indexes of mu, AG and SF are obviously improved, and the indexes are respectively improved by 52.86%, 42.38%, 44.32%, 46.57%, 39.64%, 45.68%, 9.24%, 21.72% and 5.98%. The remarkable improvement of indexes such as the average value, the average gradient, the spatial frequency and the like shows that the improvement effect of the brightness, the definition and the texture features of the image is good.
The anti-halation image fused by the improved IHS-Curvelet fusion algorithm is shown in FIG. 10, wherein the visible light image and the infrared image of the suburban road small-area halation scene are enhanced by the Adaptive Enhancement Algorithm (AEA).
The evaluation index pairs of the suburban road small-area halation fusion images are shown in table 6.
TABLE 6. suburb road small-area halation fusion image evaluation index
As can be seen from the index data in Table 6, the indexes of the fusion image mu, AG, SF and the like obtained by the adaptive enhancement algorithm of the invention are improved by 44.05%, 30.12% and 37.58% compared with HE fusion, 54.75%, 42.13% and 34.68% compared with MSR fusion, 59.63%, 23.36% and 29.75% compared with DCP fusion, and the improvement of the mean value, average gradient and spatial frequency indexes indicates that the sharpness and contrast of the fusion image are improved better.
The anti-halation image fused by the improved IHS-Curvelet fusion algorithm is shown in FIG. 11, wherein the visible light image and the infrared image of the halation scene in the suburban trunk road small region are enhanced by the Adaptive Enhancement Algorithm (AEA).
The ratio of the suburban trunk road small-area halation fusion image evaluation indexes is shown in table 7.
TABLE 7. suburb trunk road small-area halation fusion image evaluation index
As can be seen from the index data in Table 7, the AG index of the fusion image enhanced by the adaptive algorithm of the invention is improved by 75.08% compared with HE fusion, by 28.55% compared with MSR fusion and by 11.19% compared with DCP fusion, and the remarkable improvement of the average gradient index shows that the fusion image definition after the adaptive enhancement is adopted is better. Compared with HE fusion, SF and QAB/F indexes of the fusion image after ADCP enhancement are improved by 44.32 percent and 26.46 percent, and are improved by 31.35 percent and 18.53 percent compared with MSR fusion, and are improved by 5.63 percent and 11.15 percent compared with DCP fusion, and the indexes of spatial frequency and edge retention are obviously improved, which shows that the anti-halation fusion image obtained by the enhancement algorithm of the invention has richer visual characteristics, color texture and other detailed information.
The anti-halation image fused by the improved IHS-Curvelet fusion algorithm is shown in FIG. 12, wherein the visible light image and the infrared image of the large-area halation scene of the suburb road are enhanced by the Adaptive Enhancement Algorithm (AEA).
The evaluation index pairs of the suburban road large-area halation fusion images are shown in table 8.
TABLE 8 Large-area halation fusion image evaluation index for suburb community roads
As can be seen from the index data in Table 8, the indexes of the fusion image D, CEFU-IR, MIFU-VI and the like enhanced by the adaptive algorithm are respectively improved by 18.81%, 37.89%, 16.78%, 63.21%, 58.24%, 43.23%, 60.53%, 45.57% and 25.33% in sequence compared with the HE fusion image, the MSR fusion image and the DCP fusion image. The indexes such as the halation elimination degree, the cross entropy, the mutual information and the like are remarkably improved, the fact that the detail information such as the edge, the outline, the color and the like of the fused image obtained by the enhancement algorithm is richer is shown, all objective evaluation index results in the table are consistent with the anti-halation image fusion subjective visual result, and the best improvement effect of the anti-halation fused image by the self-adaptive enhancement algorithm is fully demonstrated.
The fusion anti-halation experiment and result analysis of the different night vision halation scene images show that the halation of the fusion image obtained by the self-adaptive enhancement algorithm is thoroughly eliminated, the brightness and definition of the image are obviously improved, objective index data show that the anti-halation fusion image index data obtained by the self-adaptive enhancement algorithm in different halation scenes are relatively good, the universality of the self-adaptive enhancement algorithm of the different night vision halation image provided by the invention for processing different halation scene images is fully verified, and the effectiveness of the self-adaptive enhancement algorithm of the different night vision halation image in the anti-halation fusion image and the universality in different halation scenes are also verified.
The above embodiments are merely illustrative of the principles and effects of the present invention, and it will be apparent to those skilled in the art that various changes and modifications can be made without departing from the inventive concept of the present invention, and the scope of the present invention is defined by the appended claims.
Claims (5)
1. A self-adaptive enhancement method of a heterogeneous night vision halation image is characterized by comprising the following steps: the method comprises the following steps:
step 1, determining halation critical gray value G of night vision visible light imagec；
Step 2, negating the night vision halation image, and estimating the initial transmissivity t of the negated image;
step 3, according to the initial transmissivity t and the halation critical gray value G of the imagecConstructing an adaptive transmittance function T;
when the vignetting degree of the night vision vignetting image is larger, the image transmissivity is properly reduced; according to the functional relation between a large number of night vision halation images with the halation degrees from small to large and the corresponding optimal image transmissivity, an adaptive transmissivity function T is constructed by means of least square nonlinear fitting, wherein the adaptive transmissivity function T is as follows:
in the formula, x is a pixel gray value of the night vision halation image;
in the process of enhancing the night vision halation image, the images with different halation degrees adopt corresponding transmissivity to adaptively enhance the night vision halation image; when the gray value x of the pixel of the halation image is less than the critical gray value G of the halationcThen, the initial transmittance t is used; when the pixel gray value x of the halation image is larger than or equal to the halation critical gray value GcThe transmittance is adjusted using the constructed adaptive transmittance function.
And 4, carrying out self-adaptive enhancement on the images with different halation degrees according to the self-adaptive transmittance function T.
2. The heterologous night vision halo of claim 1The self-adaptive enhancement method of the image is characterized in that: in step 1, determining a halation critical gray value G of a night vision visible light imagecThe method comprises the following steps:
step 1.1, calculating the inter-class squared difference theta according to the following formula, and obtaining an initial critical gray value k when theta is the maximum value:
in the formula, mGExpressed as the average gray value of the whole image
P1(k) For the probability of the gray level in the image, the expression is
Is the average gray value of pixels with gray levels from 0 to k
Step 1.2, calculating a halation critical gray value G according to the following formulac：
Gc＝k+s (2)
In the formula, s is a superposition coefficient.
3. The adaptive enhancement method for heterologous night vision vignetting images of claim 2, comprising: the function expression of the superposition coefficient s in the formula (2) is realized by the following steps:
(1) collecting points: determining an initial critical gray value k according to a maximum inter-class variance method by taking halo images with different halo degrees as research objects, and obtaining a superposition coefficient s by performing optimal threshold segmentation on the night vision halo image according to the superposition coefficient of the initial critical gray value; corresponding point sets (k) of a large number of halation images of different types of road vehicles from far and near and with the halation becoming smaller from small to large are collectedi，si)；
(2) Linear fitting: according to point set (k)i，si) The superposition coefficient s and the initial critical gray value k are inversely correlated, and least square normal fitting is carried out by adopting the following formula:
s＝a*k+b (3)
(3) determining the value of the coefficient: when the evaluation function L (a, b) is set to the minimum value, the values of a and b in the equation s are set to-1.11 and 181.14, respectively.
4. The adaptive enhancement method for heterologous night vision vignetting images of claim 1, comprising: in step 2, the initial transmittance t of the inverted image is estimated according to the following formula:
in the formula, omega is a defogging coefficient; i (x) is a vignetting inversion image; a is the atmospheric light intensity at infinity; avref denotes mean filtering.
5. The adaptive enhancement method for heterologous night vision vignetting images of claim 1, comprising: in step 4, the transmittance of the halation image is determined according to the adaptive transmittance function T, and the enhanced low-illuminance image j (x) is obtained by combining the following formula:
the low-intensity image j (x) is re-inverted to obtain an enhanced vignetting image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111054226.6A CN113947536A (en) | 2021-09-09 | 2021-09-09 | Self-adaptive enhancement method for heterologous night vision halation image |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111054226.6A CN113947536A (en) | 2021-09-09 | 2021-09-09 | Self-adaptive enhancement method for heterologous night vision halation image |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113947536A true CN113947536A (en) | 2022-01-18 |
Family
ID=79328298
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111054226.6A Pending CN113947536A (en) | 2021-09-09 | 2021-09-09 | Self-adaptive enhancement method for heterologous night vision halation image |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113947536A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116543378A (en) * | 2023-07-05 | 2023-08-04 | 杭州海康威视数字技术股份有限公司 | Image recognition method and device, electronic equipment and storage medium |
-
2021
- 2021-09-09 CN CN202111054226.6A patent/CN113947536A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116543378A (en) * | 2023-07-05 | 2023-08-04 | 杭州海康威视数字技术股份有限公司 | Image recognition method and device, electronic equipment and storage medium |
| CN116543378B (en) * | 2023-07-05 | 2023-09-29 | 杭州海康威视数字技术股份有限公司 | Image recognition method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110148095B (en) | Underwater image enhancement method and enhancement device | |
| CN105631829B (en) | Night haze image defogging method based on dark channel prior and color correction | |
| CN111292258B (en) | Image defogging method based on dark channel prior and bright channel prior | |
| CN106530257A (en) | Remote sensing image de-fogging method based on dark channel prior model | |
| CN109087254B (en) | Unmanned aerial vehicle aerial image haze sky and white area self-adaptive processing method | |
| CN102831591B (en) | Gaussian filter-based real-time defogging method for single image | |
| CN107945149B (en) | Automobile anti-halation method for enhancing fusion of visible light and infrared image by IHS-Curvelet transformation | |
| CN107103591A (en) | A kind of single image to the fog method based on image haze concentration sealing | |
| CN106875351A (en) | A kind of defogging method towards large area sky areas image | |
| CN112200746B (en) | Defogging method and equipment for foggy-day traffic scene image | |
| CN111861896A (en) | UUV-oriented underwater image color compensation and recovery method | |
| CN116823686B (en) | Night infrared and visible light image fusion method based on image enhancement | |
| CN110782407A (en) | Single image defogging method based on sky region probability segmentation | |
| CN111598814B (en) | Single image defogging method based on extreme scattering channel | |
| Kenk et al. | Visibility enhancer: adaptable for distorted traffic scenes by dusty weather | |
| CN114693548B (en) | Dark channel defogging method based on bright area detection | |
| CN106780362B (en) | Road video defogging method based on dichromatic reflection model and bilateral filtering | |
| CN113947536A (en) | Self-adaptive enhancement method for heterologous night vision halation image | |
| CN109345479B (en) | Real-time preprocessing method and storage medium for video monitoring data | |
| CN107437241B (en) | Dark channel image defogging method combined with edge detection | |
| CN114581914A (en) | License plate recognition method in haze environment | |
| CN105931193A (en) | Night traffic block port image enhancement method based on dark channel prior | |
| CN108416739B (en) | Traffic image defogging method based on contour wave and Markov random field | |
| CN113298730B (en) | Defogging restoration method based on image decomposition | |
| CN112270690B (en) | Self-adaptive night lane line detection method based on improved CLAHE and sliding window search |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |