CN111611907B - An image-enhanced infrared target detection method - Google Patents
An image-enhanced infrared target detection method Download PDFInfo
- Publication number
- CN111611907B CN111611907B CN202010419589.4A CN202010419589A CN111611907B CN 111611907 B CN111611907 B CN 111611907B CN 202010419589 A CN202010419589 A CN 202010419589A CN 111611907 B CN111611907 B CN 111611907B
- Authority
- CN
- China
- Prior art keywords
- image
- target detection
- infrared
- target
- enhancement
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 89
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 41
- 238000000034 method Methods 0.000 claims abstract description 28
- 238000004364 calculation method Methods 0.000 claims description 17
- 230000009466 transformation Effects 0.000 claims description 17
- 238000013527 convolutional neural network Methods 0.000 claims description 16
- 230000003044 adaptive effect Effects 0.000 claims description 15
- 238000011426 transformation method Methods 0.000 claims description 4
- 238000006243 chemical reaction Methods 0.000 abstract description 29
- 238000013528 artificial neural network Methods 0.000 abstract description 5
- 230000005855 radiation Effects 0.000 abstract description 3
- 238000000605 extraction Methods 0.000 abstract description 2
- 230000000694 effects Effects 0.000 description 9
- 238000003708 edge detection Methods 0.000 description 3
- 230000002708 enhancing effect Effects 0.000 description 3
- 238000011156 evaluation Methods 0.000 description 3
- 238000003672 processing method Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000004913 activation Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000002146 bilateral effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/20—Scenes; Scene-specific elements in augmented reality scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
-
- 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
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an image-enhanced infrared target detection method, which belongs to the technical field of infrared target attribute identification, and comprises the steps of establishing a clear infrared image data set, and setting related parameters of a target detection network, such as learning rate, momentum and the like; training a target detection algorithm network; performing self-adaptive gamma conversion on the infrared image to increase the contrast of the infrared image; utilizing a Sobel operator edge enhancement method to enhance details of the image; and inputting the infrared image subjected to the image enhancement algorithm into a target detection algorithm to extract characteristics and detect and classify. By introducing the image enhancement algorithm, the method effectively avoids the problem that the convolution neural network target detection algorithm has no obvious characteristic on infrared image extraction, and solves the problem that the detection of an infrared image target and an environment thermal radiation intensity difference or a far small target generates false recognition and missing recognition.
Description
Technical Field
The invention relates to the technical field of infrared target attribute identification, in particular to an image-enhanced infrared target detection method.
Background
The main function of infrared target detection is to locate the target of interest according to the input infrared image information, to classify the located target specifically and to give out confidence score finally; the infrared target detection technology is widely applied to the field of intelligent traffic management at present, and can remarkably improve the performance of traffic supervision and vehicle management in an intelligent traffic management system; in recent years, a great deal of research is carried out on target detection by students at home and abroad, including methods such as target detection based on pixel characteristics, target detection based on characteristic descriptors, target detection based on a gray singular value method, target detection based on a magneto-resistance sensor, a target detection method based on a BP neural network and the like, and infrared targets which are interested in people are detected by the students by using the target detection method based on the convolutional neural network, so that a very excellent effect can be achieved under certain conditions; however, under the condition that the input infrared image is not clear enough and the characteristics of the target to be identified are not obvious enough, the target detection algorithm of the convolutional neural network is easy to generate the problems of false identification and missing identification; meanwhile, many scholars have improved the target detection algorithm based on convolutional neural network, such as: the method comprises the steps of deepening the layer number of the convolutional neural network, changing the convolutional neural network activation function, combining the multi-layer feature map and the like, wherein the methods increase the diversity of extracted features, but under the condition that the features of the infrared image target to be identified are not obvious, the extracted features still can not well distinguish the target attributes and repeatedly extract irrelevant features.
For the shot infrared image, due to the limitation of shooting equipment, the quality of the infrared image is uneven, the contrast of the shot infrared image is low, and the imaging of a remote target is unclear. For gray infrared images, the contrast is critical to the visual effect, and in general, the greater the contrast, the sharper the image. The self-adaptive gamma conversion can automatically select proper gamma values according to the brightness condition of the shot infrared image, gamma conversion is carried out on the input infrared image, and stretching is carried out on dark areas or bright areas of the image to different degrees, so that the contrast of the image is enhanced, and the image can be more clear. Meanwhile, under the condition that the shot target is far or the shot target is small, the shot target has low heat radiation intensity and is not clear in imaging, and the edge detection method based on the Sobel operator is used for extracting and enhancing the edge of the input infrared image, so that the definition degree of the target is increased, the problem of target characteristic blurring caused by gray level compression caused by gamma conversion is weakened, and the image is clearer.
Currently, scholars have proposed the use of gamma conversion or Sobel operator edge detection in the field of image processing. Meanwhile, a learner uses an image enhancement algorithm to enhance the infrared image, and combines the traditional target attribute algorithm to perform target detection, so that the defect of unobvious extraction characteristics caused by unclear infrared image can be overcome, and the accuracy of target detection is improved.
Because the development time of the target detection algorithm based on the convolutional neural network is not long, theoretical basis and application popularization are further required to be studied deeply, gamma conversion is performed by adaptively selecting gamma values according to image conditions at home and abroad, and edge enhancement is performed by using a Sobel operator, and related documents are few. Meanwhile, the method is combined with a target detection algorithm based on a convolutional neural network to be applied to infrared target detection, and related researches are few.
It is therefore desirable to devise an image-enhanced infrared target detection method that overcomes or at least alleviates the above-mentioned drawbacks of the prior art.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an image enhancement infrared target detection method, which introduces an image enhancement algorithm into a convolution neural network infrared target detection algorithm, uses self-adaptive gamma transformation and enhances image contrast; meanwhile, an edge enhancement method based on a Sobel operator is used for enhancing relevant details of the image, so that the problem that false recognition and missing recognition are generated when a convolution neural network infrared target detection algorithm detects a target with small or far-small difference of the intensity of the thermal radiation of the environment is solved, and the confidence and accuracy of the convolution neural network infrared target detection are improved.
In order to solve the technical problems, the invention adopts the following technical scheme: an image-enhanced infrared target detection method, the flow of which is shown in figure 1, comprises the following steps:
step 1: acquiring a target detection image and carrying out gray level conversion according to gray level i;
step 2: performing gamma transformation on the gray level image detected by the target based on the self-adaptive gamma transformation method to obtain a transformed image s;
step 2.1: determining a gamma value of the self-adaptive gamma conversion according to the number of pixel points of each gray level;
step 2.1.1: counting the number num of pixel points corresponding to each gray level of the target detection image after gray level conversion i ；
Step 2.1.2: based on the self-adaptive gamma conversion method, the corresponding gray level gray is obtained when the number of the accumulated pixels of 15 adjacent gray levels in the target detection image is maximum, and the calculation formula is as follows:
step 2.1.3: the gamma value of the adaptive gamma transformation is determined, and the calculation formula is as follows:
step 2.2: the gamma conversion is carried out on the target detection image, and the process is as follows:
s＝cr γ ,(c,γ＞0)(s,r∈[0,1])
where r is a normalized value of the pixel gray value of the input image, s is a normalized value of the pixel gray value of the output image, and c and γ are constants.
Step 3: performing edge enhancement on the image s after the self-adaptive gamma conversion based on a Sobel operator to obtain an enhanced image G;
step 3.1: based on a Sobel operator, an X-direction edge image is obtained for the self-adaptive gamma-transformed image s, and the calculation formula is as follows:
wherein G is x Is the X-direction edge image of the s image;
step 3.2: and (3) obtaining a Y-direction edge image of the self-adaptive gamma-transformed image s, wherein the calculation formula is as follows:
wherein G is y Is the Y-direction edge image of the s-image;
step 3.3: calculating s image G after image edge enhancement, wherein the calculation formula is as follows:
G * ＝a×G+b×s，(a+b＝1)
wherein a and b are weight values.
Step 4: and inputting the image G subjected to edge enhancement into a target detection algorithm based on a convolutional neural network to detect a target.
The beneficial effects of adopting above-mentioned technical scheme to produce lie in:
(1) The self-adaptive gamma conversion for automatically selecting proper gamma value according to the brightness condition of the picture is designed on the basis of the gamma conversion, the infrared enhancement efficiency of the gamma conversion is improved, and the problem that proper gamma value needs to be manually adjusted due to different brightness conditions of the picture is effectively avoided;
(2) The Sobel operator-based edge enhancement method capable of increasing image details is designed on the basis of Sobel operator edge detection, the details of the infrared image target are improved, and the problem of gray level disappearance caused by gamma conversion is effectively avoided;
(3) The method for detecting the infrared target by image enhancement is designed, and the self-adaptive gamma transformation and the Sobel operator-based edge enhancement method are introduced into a convolutional neural network target detection algorithm, so that the definition degree of an image and the prominence of target features are improved, the problem that the features extracted by the convolutional neural network are not obvious is effectively avoided, and the confidence and the accuracy of the convolutional neural network infrared target detection are improved.
Drawings
FIG. 1 is a flow chart of an image enhanced infrared target detection method of the present invention;
FIG. 2 is a comparison chart of the detection of infrared targets before and after the method of the present invention is introduced in the embodiment of the present invention;
FIG. 3 is a graph comparing the confidence of infrared target detection before and after the method of the present invention is introduced in the embodiment of the present invention.
Detailed Description
The following describes in further detail the embodiments of the present invention with reference to the drawings and examples. The following examples are illustrative of the invention and are not intended to limit the scope of the invention.
As shown in fig. 1, the method of this embodiment is as follows.
Step 1: acquiring a target detection image and carrying out gray level conversion according to gray level i;
step 2: performing gamma transformation on the gray level image detected by the target based on the self-adaptive gamma transformation method to obtain a transformed image s;
step 2.1: determining a gamma value of the self-adaptive gamma conversion according to the number of pixel points of each gray level;
step 2.1.1: counting the number num of pixel points corresponding to each gray level of the target detection image after gray level conversion i ；
Step 2.1.2: based on the self-adaptive gamma conversion method, the corresponding gray level gray is obtained when the number of the accumulated pixels of 15 adjacent gray levels in the target detection image is maximum, and the calculation formula is as follows:
step 2.1.3: the gamma value of the adaptive gamma transformation is determined, and the calculation formula is as follows:
step 2.2: the gamma conversion is carried out on the target detection image, and the process is as follows:
s＝cr γ ,(c,γ＞0)(s,r∈[0,1])
where r is a normalized value of the pixel gray value of the input image, s is a normalized value of the pixel gray value of the output image, and c and γ are constants, in this embodiment, c=1.
Step 3: performing edge enhancement on the image s after the self-adaptive gamma conversion based on a Sobel operator to obtain an enhanced image G;
step 3.1: based on a Sobel operator, an X-direction edge image is obtained for the self-adaptive gamma-transformed image s, and the calculation formula is as follows:
wherein G is x Is the X-direction edge image of the s image;
step 3.2: and (3) obtaining a Y-direction edge image of the self-adaptive gamma-transformed image s, wherein the calculation formula is as follows:
wherein G is y Is the Y-direction edge image of the s-image;
step 3.3: calculating s image G after image edge enhancement, wherein the calculation formula is as follows:
G * ＝a×G+b×s，(a+b＝1)
in this embodiment, a=0.3 and b=0.7.
Step 4: and inputting the image G subjected to edge enhancement into a target detection algorithm based on a convolutional neural network to detect a target.
In order to verify the effectiveness and feasibility of the self-adaptive gamma transformation (namely AGT-enhancement algorithm) and the image enhanced infrared target detection method (namely AE-enhancement algorithm), a contrast test is respectively designed by using Tensorflow-Linux. For an AGT-enhancement algorithm, a single gamma value is designed to Enhance an infrared image, gamma=1, 0.5, 2 and 3.5 (namely GT-enhancement-a, b, c, d) are respectively adopted, a Mask R-CNN target detection algorithm is selected for the enhanced infrared image to detect, and the detection result is compared with the detection result enhanced by the algorithm for adaptively selecting the gamma value; for the AE-enhancement algorithm, the infrared image detection result and the infrared image detection result which are not enhanced (i.e. No-enhancement algorithm) are enhanced or denoised by adopting a common image processing method, such as global histogram equalization (i.e. GHE-enhancement algorithm), local histogram equalization (i.e. LHE-enhancement algorithm), limited contrast adaptive histogram equalization (i.e. CLAHE-enhancement algorithm), multi-scale Retinex enhancement (i.e. MSR-enhancement algorithm), bilateral filtering (i.e. BF-enhancement algorithm) and median filtering (i.e. MF-enhancement algorithm), improved adaptive gamma transformation (i.e. AGT-enhancement algorithm) and an edge enhancement method based on Sobel operator (i.e. EEBS-enhancement algorithm), the image processing method is put into enhancing the infrared image, and is detected by using a Mask R-CNN target detection method, and compared with the AE-enhancement algorithm of the invention, the experimental results are shown in tables 1 and 2:
table 1 adaptive gamma conversion experiments
As can be seen from the statistics of table 1, when γ=0.5 or 2, that is, GT-enhancement-b and GT-enhancement-c, the detection results of the infrared images after the gamma conversion are higher than those of the infrared images without the gamma conversion GT-enhancement-a, and are higher by 1.52% and 2.01% respectively in the AP values. When γ=3.5, i.e., GT-enhancement-d, the detection result of the infrared image after gamma conversion is lower than 1.65% of the detection result of GT-enhancement-a. When the adaptive gamma conversion, that is, AGT-enhancement, is adopted, the detection effect is higher than that of the gamma conversion adopting fixed values, and the detection effect is respectively 2.87%, 1.35%, 0.86% and 4.52% higher than that of GT-enhancement-a, GT-enhancement-b, GT-enhancement-c and GT-enhancement-d in AP values.
Table 2 detection results of infrared images after different enhancement algorithms
From the statistics of Table 2, it can be seen that only CLAHE-enhancement and MF-enhancement have positive effects on infrared image target detection in other image processing methods selected. The detection results are respectively 0.39% and 1.19% higher than the detection results without any enhancement method No-enhancement on the AP value, and the enhancement effect is not obvious. The detection effect of the infrared image is 2.87% higher than that of the image enhanced method No-enhancement by adopting No adaptive gamma conversion. Similarly, the detection effect of EEBS-enhancement is 2.36% higher than that of No-enhancement by only adopting the edge enhancement method based on Sobel operator. The image enhancement method AGT-enhancement of the adaptive gamma transformation has the best effect in the single image enhancement method. AE-enhancement is the infrared target detection method of the image enhancement, the detection effect is best, the AP value is 54.10%, and the detection effects are respectively 4.48% and 1.61% higher than those of No-enhancement and AGT-enhancement.
The AP value is an evaluation index of the target detection algorithm, and in this embodiment, an evaluation system in the COCO challenge is used: AP, ap@50, ap@75. The calculation formula of the AP value is as follows:
wherein Precision is the Precision, recall is the Recall, m is the number of images detected in the detection result, and i is the ith detected image; TP is the detection result of correctly detecting the foreground object as the foreground object, FP is the detection result of incorrectly detecting the background as the foreground object, TN is the detection result of correctly detecting the background as the background, and FN is the detection result of incorrectly detecting the foreground object as the background.
In the target detection algorithm, ioU (Intersection-over-Union) is a threshold for evaluating whether the detection algorithm recognizes a foreground target, i.e. determining that the foreground target is Positive when the detection confidence of the foreground target is higher than IoU threshold. Therefore, different IoU is set, and the calculated number of Positive samples and the calculated AP value are different, and the calculation formula of IoU is as follows:
wherein, in the method, ,
wherein S is A To mark the area of the frame (group Truth), S B To predict the area of the positioning frame.
(1) AP: in the present embodiment, the main evaluation index, when IoU is in the range of 0.5 to 0.95, the AP value is calculated every 0.05,
and average all AP values;
(2) Ap@50: AP value when IoU takes 0.5;
(3) Ap@75: AP value when IoU takes 0.75;
(4) AP@S: for the number of pixel points less than 32 2 The AP value in (1) is calculated for the target;
(5) AP@M: for the number of pixels greater than 32 2 And less than 96 2 The AP value in (1) is calculated for the target;
(6) AP@L: for pixel point number greater than 96 2 The AP value in (1) is calculated for the target.
As can be seen from fig. 2 and 3, the missing recognition phenomenon of the AE-enhancement algorithm is significantly reduced compared with that of the No-enhancement algorithm, and the confidence of detection is significantly improved. The detection confidence and the recognition accuracy are effectively improved due to the missing recognition phenomenon of the AE-enhancement algorithm infrared target detection, and a novel method and a novel approach are provided for the problem of infrared target detection based on a convolutional neural network.
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010419589.4A CN111611907B (en) | 2020-05-18 | 2020-05-18 | An image-enhanced infrared target detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010419589.4A CN111611907B (en) | 2020-05-18 | 2020-05-18 | An image-enhanced infrared target detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111611907A CN111611907A (en) | 2020-09-01 |
| CN111611907B true CN111611907B (en) | 2023-10-31 |
Family
ID=72204982
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010419589.4A Active CN111611907B (en) | 2020-05-18 | 2020-05-18 | An image-enhanced infrared target detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111611907B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112287772B (en) * | 2020-10-10 | 2023-02-10 | 深圳市中达瑞和科技有限公司 | Fingerprint trace detection method, fingerprint detection device and computer readable storage medium |
| CN112200788B (en) * | 2020-10-16 | 2021-03-26 | 清华大学 | High-temperature deformation measuring device and method |
| CN112365426B (en) * | 2020-11-25 | 2022-06-07 | 兰州理工大学 | Infrared image edge enhancement method based on double-branch convolution neural network |
| CN112700386A (en) * | 2020-12-30 | 2021-04-23 | 江苏奥格视特信息科技有限公司 | Fuzzy image ultrahigh-definition transformation method and system based on artificial intelligence |
| CN115330644B (en) * | 2022-10-13 | 2023-04-07 | 极限人工智能有限公司 | Adaptive edge enhancement method and system for endoscope image |
| CN116030052B (en) * | 2023-03-27 | 2023-06-16 | 惠州威尔高电子有限公司 | Etching quality detection method for lamination process of computer display panel |
| CN118644723B (en) * | 2024-06-18 | 2024-12-06 | 广州市星飞达电子科技有限公司 | Infrared image target detection method and system based on deep learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104574435A (en) * | 2014-09-24 | 2015-04-29 | 中国人民解放军国防科学技术大学 | Motion camera foreground segmentation method based on block clustering |
| CN110599415A (en) * | 2019-08-29 | 2019-12-20 | 西安电子科技大学 | Image contrast enhancement implementation method based on local adaptive gamma correction |
| CN110930390A (en) * | 2019-11-22 | 2020-03-27 | 郑州智利信信息技术有限公司 | Chip pin missing detection method based on semi-supervised deep learning |
-
2020
- 2020-05-18 CN CN202010419589.4A patent/CN111611907B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104574435A (en) * | 2014-09-24 | 2015-04-29 | 中国人民解放军国防科学技术大学 | Motion camera foreground segmentation method based on block clustering |
| CN110599415A (en) * | 2019-08-29 | 2019-12-20 | 西安电子科技大学 | Image contrast enhancement implementation method based on local adaptive gamma correction |
| CN110930390A (en) * | 2019-11-22 | 2020-03-27 | 郑州智利信信息技术有限公司 | Chip pin missing detection method based on semi-supervised deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111611907A (en) | 2020-09-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111611907B (en) | An image-enhanced infrared target detection method | |
| CN104408707B (en) | Rapid digital imaging fuzzy identification and restored image quality assessment method | |
| CN110490914A (en) | It is a kind of based on brightness adaptively and conspicuousness detect image interfusion method | |
| CN111325721A (en) | Gas leakage detection method and system based on infrared thermal imaging | |
| CN107909081B (en) | A fast acquisition and fast calibration method for image datasets in deep learning | |
| CN107169985A (en) | A kind of moving target detecting method based on symmetrical inter-frame difference and context update | |
| CN108765491A (en) | A kind of SAR image Ship Target Detection method | |
| CN102768731A (en) | System and method for automatic positioning and recognition of targets based on high-definition video images | |
| CN102799862A (en) | System and method for pedestrian rapid positioning and event detection based on high definition video monitor image | |
| CN106203334A (en) | A method for flame detection in indoor scenes | |
| CN114926407A (en) | Steel surface defect detection system based on deep learning | |
| CN103903018A (en) | Method and system for positioning license plate in complex scene | |
| CN108804992B (en) | Crowd counting method based on deep learning | |
| CN102902951A (en) | System and method for vehicle target location and event detection on basis of high-definition video monitoring images | |
| CN109242032B (en) | A target detection method based on deep learning | |
| CN109190455B (en) | Smoky vehicle identification method based on Gaussian mixture and autoregressive moving average model | |
| CN111079688A (en) | Living body detection method based on infrared image in face recognition | |
| CN110210316A (en) | Traffic lights digit recognition method based on gray level image | |
| CN116665015B (en) | A method for detecting weak and small targets in infrared sequence images based on YOLOv5 | |
| CN115512224A (en) | A Curve Fitting Contrast Enhanced Road Crack Detection Method | |
| CN112419261B (en) | Visual acquisition method and device with abnormal point removing function | |
| CN113066077B (en) | Flame detection method and device | |
| CN113221603A (en) | Method and device for detecting shielding of monitoring equipment by foreign matters | |
| CN117522778A (en) | Hollow brick flaw detection system | |
| CN117197064A (en) | Automatic non-contact eye red degree analysis method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |