CN113705501B - Marine target detection method and system based on image recognition technology - Google Patents
Marine target detection method and system based on image recognition technology Download PDFInfo
- Publication number
- CN113705501B CN113705501B CN202111025400.4A CN202111025400A CN113705501B CN 113705501 B CN113705501 B CN 113705501B CN 202111025400 A CN202111025400 A CN 202111025400A CN 113705501 B CN113705501 B CN 113705501B
- Authority
- CN
- China
- Prior art keywords
- image
- backlight
- original image
- gray level
- preset
- 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 49
- 238000005516 engineering process Methods 0.000 title claims abstract description 15
- 238000007781 pre-processing Methods 0.000 claims abstract description 46
- 238000000034 method Methods 0.000 claims abstract description 23
- 238000006243 chemical reaction Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 8
- 230000009466 transformation Effects 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 4
- 230000001105 regulatory effect Effects 0.000 claims description 4
- 238000003062 neural network model Methods 0.000 claims description 3
- 230000009286 beneficial effect Effects 0.000 abstract 1
- 230000008569 process Effects 0.000 description 6
- 230000002093 peripheral effect Effects 0.000 description 4
- 230000011218 segmentation Effects 0.000 description 4
- 238000003709 image segmentation Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Processing (AREA)
Abstract
The invention provides a method and a system for detecting an offshore target based on an image recognition technology, and in one aspect, the method comprises the following steps of S1, acquiring an original image to be detected of the offshore target; s2, judging whether the original image is a backlight image or not; s3, if the original image is a backlight image, preprocessing the original image by adopting a preset backlight image preprocessing algorithm to obtain a preprocessed image; if the original image is a non-backlight image, preprocessing the original image by adopting a preset forward-light image preprocessing algorithm to obtain a preprocessed image; s4, detecting the offshore targets based on the preprocessed images to obtain detection results. On the other hand, the invention also provides a system for realizing the method. The invention effectively enhances the pertinence of preprocessing the backlight image, thereby improving the accuracy of the preprocessed image obtained by the invention, and further being beneficial to improving the accuracy of offshore target detection.
Description
Technical Field
The invention relates to the field of detection, in particular to a marine target detection method and system based on an image recognition technology.
Background
In the prior art, when an image recognition technology is used for detecting an offshore target, the image is generally directly preprocessed, characteristic information is extracted, and then target recognition is performed based on the characteristic information. However, in the prior art, the back light scene is not considered in the preprocessing process, and under the back light condition, the pixel value left on the image of the offshore target may be smaller than that of the reflected wave, so that the influence of the wave is easy to cause the final target detection result to be inaccurate.
Disclosure of Invention
In view of the above problems, an object of the present invention is to provide a method and a system for detecting an offshore object based on an image recognition technology.
In one aspect, the invention provides a marine target detection method based on an image recognition technology, which comprises the following steps:
s1, acquiring an original image to be subjected to offshore target detection;
s2, judging whether the original image is a backlight image or not;
S3, if the original image is a backlight image, preprocessing the original image by adopting a preset backlight image preprocessing algorithm to obtain a preprocessed image;
If the original image is a non-backlight image, preprocessing the original image by adopting a preset forward-light image preprocessing algorithm to obtain a preprocessed image;
s4, detecting the offshore targets based on the preprocessed images to obtain detection results.
Preferably, the determining whether the original image is a backlight image includes:
converting the original image into a gray scale image;
Acquiring a gray level histogram of a gray level image;
carrying out normalization processing on the gray level histogram to obtain a normalized histogram S, S= (S 1,s2,…,s256);
calculating standard deviation of the normalized histogram S:
Wherein dev st (S) represents the standard deviation of the normalized histogram S, S t represents the total number of pixel points included in the t-th gray level;
Sea-sky-line detection is carried out on the gray level image, and the gray level image is divided into a sky area image and a sea surface area image based on the sea-sky-line;
Converting the sky area image into a binary image;
Detecting the connected domain of the binary image, and obtaining the total number num max of pixel points contained in the connected domain with the largest average pixel value in the binary image;
Calculating the proportion of pixel points between the connected domain with the maximum average pixel value and the sky image:
Wherein, prop represents the proportion of the pixel points between the maximum connected domain and the sky image, num sky represents the total number of the pixel points contained in the sky image;
Calculating the backlight index of the original image:
Wherein bklidx represents a backlight index of an original image, stdev st represents a standard value of standard deviation of a preset normalized histogram, prop st represents a preset proportional standard value, C represents a preset constant coefficient, α and β represent preset weight parameters, α+β=1;
judging bklidx whether the original image is larger than a preset backlight index judgment threshold, if yes, indicating that the original image is a backlight image, and if not, indicating that the original image is not the backlight image.
Preferably, the preprocessing the original image by using a preset backlight image preprocessing algorithm to obtain a preprocessed image includes:
converting the original image into a gray scale image;
carrying out gray level reversal treatment on the gray level image by adopting a conversion filter to obtain a reversed image;
And carrying out segmentation processing on the reversed image by adopting an image segmentation algorithm to obtain a preprocessed image.
Preferably, the preprocessing the original image by using a preset forward light image preprocessing algorithm to obtain a preprocessed image includes:
converting the original image into a gray scale image;
Acquiring an regional extreme point image in the gray level image;
Adjusting the regional extreme points to obtain an adjusted regional extreme point image;
and carrying out enhancement processing on the gray level image based on the regulated regional extreme point image to obtain a preprocessed image.
Preferably, the marine target detection based on the preprocessed image, to obtain a detection result, includes:
And inputting the preprocessed image into a pre-trained neural network model for recognition to obtain a detection result.
On the other hand, the invention also provides an offshore target detection system based on the image recognition technology, which comprises an acquisition module, a judgment module, a preprocessing module and a detection module;
the acquisition module is used for acquiring an original image to be subjected to offshore target detection;
the judging module is used for judging whether the original image is a backlight image or not;
the acquisition module is used for preprocessing the original image by adopting a preset backlight image preprocessing algorithm when the original image is a backlight image, so as to obtain a preprocessed image;
The method comprises the steps that when an original image is a non-backlight image, a preset forward-light image preprocessing algorithm is adopted to preprocess the original image, and a preprocessed image is obtained;
The detection module is used for detecting the offshore target based on the preprocessed image to obtain a detection result.
According to the invention, by carrying out backlight judgment on the original image and then carrying out image pretreatment on the backlight image and the non-backlight image in different pretreatment modes, the pertinence of pretreatment on the backlight image is effectively enhanced, so that the accuracy of the pretreated image obtained by the method is improved, and the accuracy of detection on an offshore target is further improved.
Drawings
The invention will be further described with reference to the accompanying drawings, in which embodiments do not constitute any limitation of the invention, and other drawings can be obtained by one of ordinary skill in the art without inventive effort from the following drawings.
FIG. 1 is a diagram of an exemplary embodiment of a method for marine object detection based on image recognition technology according to the present invention.
Detailed Description
Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative only and are not to be construed as limiting the invention.
The invention provides a marine target detection method and system based on an image recognition technology.
In one aspect, as shown in fig. 1, the present invention provides a method for detecting an offshore object based on an image recognition technology, including:
s1, acquiring an original image to be subjected to offshore target detection;
s2, judging whether the original image is a backlight image or not;
S3, if the original image is a backlight image, preprocessing the original image by adopting a preset backlight image preprocessing algorithm to obtain a preprocessed image;
If the original image is a non-backlight image, preprocessing the original image by adopting a preset forward-light image preprocessing algorithm to obtain a preprocessed image;
s4, detecting the offshore targets based on the preprocessed images to obtain detection results.
According to the invention, by carrying out backlight judgment on the original image and then carrying out image pretreatment on the backlight image and the non-backlight image in different pretreatment modes, the pertinence of pretreatment on the backlight image is effectively enhanced, so that the accuracy of the pretreated image obtained by the method is improved, and the accuracy of detection on an offshore target is further improved.
Preferably, the determining whether the original image is a backlight image includes:
converting the original image into a gray scale image;
Acquiring a gray level histogram of a gray level image;
carrying out normalization processing on the gray level histogram to obtain a normalized histogram S, S= (S 1,s2,…,s256);
calculating standard deviation of the normalized histogram S:
Wherein dev st (S) represents the standard deviation of the normalized histogram S, S t represents the total number of pixel points included in the t-th gray level;
Sea-sky-line detection is carried out on the gray level image, and the gray level image is divided into a sky area image and a sea surface area image based on the sea-sky-line;
Converting the sky area image into a binary image;
Detecting the connected domain of the binary image, and obtaining the total number num max of pixel points contained in the connected domain with the largest average pixel value in the binary image;
Calculating the proportion of pixel points between the connected domain with the maximum average pixel value and the sky image:
Wherein, prop represents the proportion of the pixel points between the maximum connected domain and the sky image, num sky represents the total number of the pixel points contained in the sky image;
Calculating the backlight index of the original image:
Wherein bklidx represents a backlight index of an original image, stdev st represents a standard value of standard deviation of a preset normalized histogram, prop st represents a preset proportional standard value, C represents a preset constant coefficient, α and β represent preset weight parameters, α+β=1;
judging bklidx whether the original image is larger than a preset backlight index judgment threshold, if yes, indicating that the original image is a backlight image, and if not, indicating that the original image is not the backlight image.
When judging the backlight image, the accurate backlight judgment result can be comprehensively obtained mainly by considering the pixel value distribution of the gray level histogram and the connected domain with the maximum average pixel value of the sky area. In the case of backlight, the pixel value of the pixel point in the sky area is generally larger than that in the sea area, so that the variance of the histogram is relatively large, and thus whether or not the backlight is performed can be determined according to the variance, but if the histogram distribution is considered only, the influence of extreme conditions, such as when the area occupied by the sun in the original image is relatively small, the standard deviation of the histogram is not large even in the case of backlight, and there is a risk of erroneous determination. Therefore, the invention also considers the connected domain of the sky image, and because the connected domain where the sun is positioned is the connected domain with the largest average pixel value, the degree of backlight can be judged by the proportion of the pixel points between the connected domain and the sky image, and the larger the proportion is, the more backlight is, so that whether the image is backlit or not can be accurately judged.
Preferably, the preprocessing the original image by using a preset backlight image preprocessing algorithm to obtain a preprocessed image includes:
converting the original image into a gray scale image;
carrying out gray level reversal treatment on the gray level image by adopting a conversion filter to obtain a reversed image;
And carrying out segmentation processing on the reversed image by adopting an image segmentation algorithm to obtain a preprocessed image.
In backlight shooting, the pixel points of the target area are often smaller than the pixel values of the pixel points of the sea wave of the reflected sunlight, and thus, cause detection difficulty. Therefore, through setting the gray inversion process, the pixel with the original gray value smaller is converted into the pixel with the gray value larger, and the pixel with the original gray value larger is converted into the pixel with the gray value smaller, so that the influence degree of the pixel of the target area by the pixel of the sea wave reflecting the sunlight can be effectively reduced. The inversion image is segmented, so that the number of pixels participating in the subsequent target detection process is reduced, and the recognition speed of the invention is improved.
Preferably, the converting the original image into a gray scale image includes:
the original image is converted into a gray scale image by a weighted average method.
Preferably, the performing gray inversion processing on the gray image by using a transform filter to obtain an inverted image includes:
The transformation filter is adopted to be constructed by the following construction functions:
Wherein x and y represent an abscissa and an ordinate of a pixel point in a gray scale image, δ 1 represents a first inversion parameter, δ 2 represents a second inversion parameter ,δ1∈(c1+0.16h,c1+0.17h),δ2＝c2-c3exp(-3.8)h,c1、c2、c3 respectively represents a preset first constant coefficient, second constant coefficient and third constant coefficient, and h represents the number of rows of the pixel point contained in the gray scale image; the value range of c 1 is 0.49,0.59, the value range of c 2 is 2.69,2.81, and the value range of c 3 is 5.99,6.19; cvp (x, y) represents a build function;
Constructing a transform filter of size Q×Q using the above construction function;
the inverse image is obtained by convolving the gray image with a transform filter.
The process of the conversion filter constructed by the invention is the same as the process of the Gaussian filter, but the construction function is different, the conversion filter constructed by the construction function can press a highlight region, and meanwhile, the pixel value of the pixel point of a dark region is improved, so that the pixel value distribution of the pixel point is more uniform, and the content of detail information of a reverse image is improved. Meanwhile, the filter constructed by the invention has a certain filtering function, and can effectively reduce noise in the reverse image.
Preferably, the image segmentation algorithm is used to segment the inverted image to obtain a preprocessed image, and the method includes:
dividing the reversed image by using a mean iterative segmentation algorithm to obtain a foreground image and a background image;
Acquiring a set S of corresponding pixel points of the foreground image in the reverse image;
the pixels in set S are used to compose the preprocessed image.
In addition to the mean iterative segmentation algorithm, other algorithms that can achieve foreground and background separation, such as region growing algorithms, etc., are also possible. Because the reverse image is generated on the basis of the gray level image, each pixel point in the reverse image can find the pixel point with the same relative coordinates in the gray level image, and the foreground image can find the corresponding pixel point in the gray level image.
Preferably, the preprocessing the original image by using a preset forward light image preprocessing algorithm to obtain a preprocessed image includes:
converting the original image into a gray scale image;
Acquiring an regional extreme point image in the gray level image;
Adjusting the regional extreme points to obtain an adjusted regional extreme point image;
and carrying out enhancement processing on the gray level image based on the regulated regional extreme point image to obtain a preprocessed image.
According to the embodiment of the invention, the gray level image is enhanced by constructing the regional extreme point image, so that the difference between the target to be detected and the peripheral pixel points can be enhanced, the influence of waves on the subsequent target identification can be effectively reduced, and the noise reduction effect is realized.
Preferably, the acquiring the regional extreme point image in the gray scale image includes:
the computational convolution template is constructed as follows:
Wherein, lcb represents a convolution template, A×A represents the size of a convolution window, b represents a preset constant coefficient, and b is less than A;
and carrying out convolution calculation on each pixel point in the gray level image by using the lcb to obtain a convolution result:
jc＝lcb*gray
Where jc represents the convolution result, gray represents the gray image, and x represents the convolution operation;
Judging whether the convolution result jc is larger than a preset convolution result threshold value, if so, acquiring all the regional extreme points from the gray image by taking the pixel points corresponding to the convolution result as regional extreme points, and forming a regional extreme point image.
The purpose of the regional extreme point image is to select the pixel point with the largest difference with the peripheral pixel points, because the difference between the pixel points of the target region and the peripheral pixel points is larger in the gray level image generally, but the pixel points of the wave region also have the characteristics, and the difference between the pixel points of the wave region and the peripheral pixel points is generally smaller than the target region, so that the accuracy of obtaining the pixel points belonging to the target region in the regional extreme point image can be further improved by setting the threshold value.
Preferably, the adjusting the regional extreme point to obtain an adjusted regional extreme point image includes:
The regional extreme points are adjusted by adopting the following modes:
Wherein q' (x, y) represents a pixel value of the region extremum point at the coordinate (x, y) after adjustment, q (x, y) represents a pixel value of the region extremum point at the coordinate (x, y), q mi and q ma respectively represent a minimum value and a maximum value of the region extremum point pixel value in a k×k-sized neighborhood of the region extremum point at the coordinate (x, y), B represents a preset adjustment parameter, sh represents a valued function, and if q (x, y) -q ma is equal to 0, the value of sh (q (x, y) -q ma) is 1, otherwise, the value of sh (q (x, y) -q ma) is-1.
The purpose of the adjustment processing is to consider that the average pixel value of the pixel points of the target area is possibly lower than that of the pixel points of the wave area, so that the difference between the pixel points of the wave area and the surrounding pixel points is generally smaller than that of the target area through the adjustment processing, and the difference between the pixel points of the target area and the surrounding pixel points can be further improved after the adjustment processing, so that the subsequent accurate identification of the offshore target is facilitated.
Preferably, the enhancing the gray image based on the adjusted regional extremum point image to obtain a preprocessed image comprises:
the gray level image is enhanced by adopting the following method:
afgray(x,y)＝BL(x,y)×gray(x,y)
Wherein gray (x, y) represents a pixel value at coordinates (x, y) in the gray image, BL represents an enhancement coefficient, afgray (x, y) represents a pixel value after enhancement processing of gray (x, y),
BL (x, y) represents an enhancement coefficient, and is obtained by:
If it is Less than 1, then/>If/>Then/>If/>Greater than 2, BL (x, y) is equal to 1, where h 1、h2、h3 represents a preset scaling factor, h 1+h2+h3 =1.
In the enhancement process, since the enhancement coefficient is related to the regional extreme point image, the pixel point related to the target region can be further enhanced, thereby increasing the difference between the pixel point of the target region and the sea wave pixel point.
Preferably, the marine target detection based on the preprocessed image, to obtain a detection result, includes:
And inputting the preprocessed image into a pre-trained neural network model for recognition to obtain a detection result.
On the other hand, the invention also provides an offshore target detection system based on the image recognition technology, which comprises an acquisition module, a judgment module, a preprocessing module and a detection module;
the acquisition module is used for acquiring an original image to be subjected to offshore target detection;
the judging module is used for judging whether the original image is a backlight image or not;
the acquisition module is used for preprocessing the original image by adopting a preset backlight image preprocessing algorithm when the original image is a backlight image, so as to obtain a preprocessed image;
The method comprises the steps that when an original image is a non-backlight image, a preset forward-light image preprocessing algorithm is adopted to preprocess the original image, and a preprocessed image is obtained;
The detection module is used for detecting the offshore target based on the preprocessed image to obtain a detection result.
While embodiments of the invention have been shown and described, it will be understood by those skilled in the art that: many changes, modifications, substitutions and variations may be made to the embodiments without departing from the spirit and principles of the invention, the scope of which is defined by the claims and their equivalents.
Claims (3)
1. An offshore target detection method based on an image recognition technology is characterized by comprising the following steps:
s1, acquiring an original image to be subjected to offshore target detection;
s2, judging whether the original image is a backlight image or not;
S3, if the original image is a backlight image, preprocessing the original image by adopting a preset backlight image preprocessing algorithm to obtain a preprocessed image;
If the original image is a non-backlight image, preprocessing the original image by adopting a preset forward-light image preprocessing algorithm to obtain a preprocessed image;
S4, detecting an offshore target based on the preprocessed image to obtain a detection result;
the judging whether the original image is a backlight image or not includes:
converting the original image into a gray scale image;
Acquiring a gray level histogram of a gray level image;
carrying out normalization processing on the gray level histogram to obtain a normalized histogram S, S= (S 1,s2,…,s256);
calculating standard deviation of the normalized histogram S:
Wherein dev st (S) represents the standard deviation of the normalized histogram S, S t represents the total number of pixel points included in the t-th gray level;
Sea-sky-line detection is carried out on the gray level image, and the gray level image is divided into a sky area image and a sea surface area image based on the sea-sky-line;
Converting the sky area image into a binary image;
Detecting the connected domain of the binary image, and obtaining the total number num max of pixel points contained in the connected domain with the largest average pixel value in the binary image;
Calculating the proportion of pixel points between the connected domain with the maximum average pixel value and the sky area image:
Wherein, prop represents the proportion of the pixel points between the connected domain with the maximum average pixel value and the sky area image, num sky represents the total number of the pixel points contained in the sky area image;
Calculating the backlight index of the original image:
Wherein bklidx represents a backlight index of an original image, stdev st represents a standard value of standard deviation of a preset normalized histogram, prop st represents a preset proportional standard value, C represents a preset constant coefficient, α and β represent preset weight parameters, α+β=1;
Judging bklidx whether the original image is larger than a preset backlight index judgment threshold, if yes, indicating that the original image is a backlight image, and if not, indicating that the original image is not the backlight image;
The preprocessing of the original image by adopting a preset forward light image preprocessing algorithm to obtain a preprocessed image comprises the following steps:
converting the original image into a gray scale image;
Acquiring an regional extreme point image in the gray level image;
Adjusting the regional extreme points to obtain an adjusted regional extreme point image;
carrying out enhancement processing on the gray level image based on the regulated regional extreme point image to obtain a preprocessed image; the preprocessing of the original image by a preset backlight image preprocessing algorithm to obtain a preprocessed image comprises the following steps:
converting the original image into a gray scale image;
carrying out gray level reversal treatment on the gray level image by adopting a conversion filter to obtain a reversed image;
Dividing the reversed image by adopting an image dividing algorithm to obtain a preprocessed image;
The method for carrying out gray inversion processing on the gray image by adopting the conversion filter to obtain an inverted image comprises the following steps:
The transformation filter is adopted to be constructed by the following construction functions:
Wherein x and y represent an abscissa and an ordinate of a pixel point in a gray scale image, δ 1 represents a first inversion parameter, δ 2 represents a second inversion parameter ,δ1∈(c1+0.16h,c1+0.17h),δ2＝c2-c3exp(-3.8)h,c1、c2、c3 respectively represents a preset first constant coefficient, second constant coefficient and third constant coefficient, and h represents the number of rows of the pixel point contained in the gray scale image; the value range of c 1 is 0.49,0.59, the value range of c 2 is 2.69,2.81, and the value range of c 3 is 5.99,6.19; cvp (x, y) represents a build function;
Constructing a transform filter of size Q×Q using the above construction function;
and carrying out convolution processing on the gray level image by using a transformation filter to obtain a reverse image.
2. The method for detecting an offshore target based on the image recognition technology according to claim 1, wherein the step of performing the offshore target detection based on the preprocessed image to obtain the detection result comprises the steps of:
And inputting the preprocessed image into a pre-trained neural network model for recognition to obtain a detection result.
3. The marine target detection system based on the image recognition technology is characterized by comprising an acquisition module, a judgment module, a preprocessing module and a detection module;
the acquisition module is used for acquiring an original image to be subjected to offshore target detection;
the judging module is used for judging whether the original image is a backlight image or not;
the acquisition module is used for preprocessing the original image by adopting a preset backlight image preprocessing algorithm when the original image is a backlight image, so as to obtain a preprocessed image;
The method comprises the steps that when an original image is a non-backlight image, a preset forward-light image preprocessing algorithm is adopted to preprocess the original image, and a preprocessed image is obtained;
the detection module is used for detecting the offshore target based on the preprocessed image to obtain a detection result;
the judging whether the original image is a backlight image or not includes:
converting the original image into a gray scale image;
Acquiring a gray level histogram of a gray level image;
carrying out normalization processing on the gray level histogram to obtain a normalized histogram S, S= (S 1,s2,…,s256);
calculating standard deviation of the normalized histogram S:
Wherein dev st (S) represents the standard deviation of the normalized histogram S, S t represents the total number of pixel points included in the t-th gray level;
Sea-sky-line detection is carried out on the gray level image, and the gray level image is divided into a sky area image and a sea surface area image based on the sea-sky-line;
Converting the sky area image into a binary image;
Detecting the connected domain of the binary image, and obtaining the total number num max of pixel points contained in the connected domain with the largest average pixel value in the binary image;
Calculating the proportion of pixel points between the connected domain with the maximum average pixel value and the sky area image:
Wherein, prop represents the proportion of the pixel points between the connected domain with the maximum average pixel value and the sky area image, num sky represents the total number of the pixel points contained in the sky area image;
Calculating the backlight index of the original image:
Wherein bklidx represents a backlight index of an original image, stdev st represents a standard value of standard deviation of a preset normalized histogram, prop st represents a preset proportional standard value, C represents a preset constant coefficient, α and β represent preset weight parameters, α+β=1;
Judging bklidx whether the original image is larger than a preset backlight index judgment threshold, if yes, indicating that the original image is a backlight image, and if not, indicating that the original image is not the backlight image;
The preprocessing of the original image by adopting a preset forward light image preprocessing algorithm to obtain a preprocessed image comprises the following steps:
converting the original image into a gray scale image;
Acquiring an regional extreme point image in the gray level image;
Adjusting the regional extreme points to obtain an adjusted regional extreme point image;
carrying out enhancement processing on the gray level image based on the regulated regional extreme point image to obtain a preprocessed image;
the preprocessing of the original image by a preset backlight image preprocessing algorithm to obtain a preprocessed image comprises the following steps:
converting the original image into a gray scale image;
carrying out gray level reversal treatment on the gray level image by adopting a conversion filter to obtain a reversed image;
Dividing the reversed image by adopting an image dividing algorithm to obtain a preprocessed image;
The method for carrying out gray inversion processing on the gray image by adopting the conversion filter to obtain an inverted image comprises the following steps:
The transformation filter is adopted to be constructed by the following construction functions:
Wherein x and y represent an abscissa and an ordinate of a pixel point in a gray scale image, δ 1 represents a first inversion parameter, δ 2 represents a second inversion parameter ,δ1∈(c1+0.16h,c1+0.17h),δ2＝c2-c3exp(-3.8)h,c1、c2、c3 respectively represents a preset first constant coefficient, second constant coefficient and third constant coefficient, and h represents the number of rows of the pixel point contained in the gray scale image; the value range of c 1 is 0.49,0.59, the value range of c 2 is 2.69,2.81, and the value range of c 3 is 5.99,6.19; cvp (x, y) represents a build function;
Constructing a transform filter of size Q×Q using the above construction function;
and carrying out convolution processing on the gray level image by using a transformation filter to obtain a reverse image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111025400.4A CN113705501B (en) | 2021-09-02 | 2021-09-02 | Marine target detection method and system based on image recognition technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111025400.4A CN113705501B (en) | 2021-09-02 | 2021-09-02 | Marine target detection method and system based on image recognition technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113705501A CN113705501A (en) | 2021-11-26 |
| CN113705501B true CN113705501B (en) | 2024-04-26 |
Family
ID=78658898
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111025400.4A Active CN113705501B (en) | 2021-09-02 | 2021-09-02 | Marine target detection method and system based on image recognition technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113705501B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114219992B (en) * | 2021-12-14 | 2022-06-03 | 杭州古伽船舶科技有限公司 | Unmanned ship obstacle avoidance system based on image recognition technology |
| CN116758508B (en) * | 2023-08-18 | 2024-01-12 | 四川蜀道新能源科技发展有限公司 | Pavement marking detection method, system and terminal based on pixel difference expansion processing |
Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101604383A (en) * | 2009-07-24 | 2009-12-16 | 哈尔滨工业大学 | A kind of method for detecting targets at sea based on infrared image |
| CN102542552A (en) * | 2010-12-21 | 2012-07-04 | 北京汉王智通科技有限公司 | Frontlighting and backlighting judgment of video images and detection method of shooting time |
| KR101172283B1 (en) * | 2012-05-31 | 2012-08-09 | 국방과학연구소 | Method for detecting maritime targets |
| CN104408974A (en) * | 2014-07-15 | 2015-03-11 | 王晓东 | Ship peripheral target detection alarm system |
| EP2905722A1 (en) * | 2014-02-10 | 2015-08-12 | Huawei Technologies Co., Ltd. | Method and apparatus for detecting salient region of image |
| CN106161967A (en) * | 2016-09-13 | 2016-11-23 | 维沃移动通信有限公司 | A kind of backlight scene panorama shooting method and mobile terminal |
| CN106570487A (en) * | 2016-11-10 | 2017-04-19 | 维森软件技术（上海）有限公司 | Method and device for predicting collision between objects |
| KR101800591B1 (en) * | 2017-10-30 | 2017-11-24 | 김용오 | System for vehicle number recognition using 3D camera |
| JP6336693B1 (en) * | 2016-12-09 | 2018-06-06 | 株式会社日立国際電気 | Water intrusion detection system and method |
| CN109101897A (en) * | 2018-07-20 | 2018-12-28 | 中国科学院自动化研究所 | Object detection method, system and the relevant device of underwater robot |
| WO2019011147A1 (en) * | 2017-07-10 | 2019-01-17 | Oppo广东移动通信有限公司 | Human face region processing method and apparatus in backlight scene |
| CN109242870A (en) * | 2018-07-13 | 2019-01-18 | 上海大学 | A kind of sea horizon detection method divided based on image with textural characteristics |
| CN109978869A (en) * | 2019-03-29 | 2019-07-05 | 清华大学 | A kind of sea horizon detection method and system based on gray level co-occurrence matrixes and Hough transform |
| CN109993744A (en) * | 2019-04-09 | 2019-07-09 | 大连海事大学 | A kind of infrared target detection method under sea backlight environment |
| WO2019233393A1 (en) * | 2018-06-08 | 2019-12-12 | Oppo广东移动通信有限公司 | Image processing method and apparatus, storage medium, and electronic device |
| CN111489344A (en) * | 2020-04-10 | 2020-08-04 | 湖南索莱智能科技有限公司 | Method, system and related device for determining image definition |
| CN111898633A (en) * | 2020-06-19 | 2020-11-06 | 北京理工大学 | High-spectral image-based marine ship target detection method |
| CN112085753A (en) * | 2020-09-02 | 2020-12-15 | 广东海启星海洋科技有限公司 | Water level monitoring method, equipment, medium and monitoring system based on image processing |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP4956452B2 (en) * | 2008-01-25 | 2012-06-20 | 富士重工業株式会社 | Vehicle environment recognition device |
| CN103217436B (en) * | 2013-03-06 | 2015-05-20 | 京东方科技集团股份有限公司 | Backlight module group defect detection method and equipment |
-
2021
- 2021-09-02 CN CN202111025400.4A patent/CN113705501B/en active Active
Patent Citations (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101604383A (en) * | 2009-07-24 | 2009-12-16 | 哈尔滨工业大学 | A kind of method for detecting targets at sea based on infrared image |
| CN102542552A (en) * | 2010-12-21 | 2012-07-04 | 北京汉王智通科技有限公司 | Frontlighting and backlighting judgment of video images and detection method of shooting time |
| KR101172283B1 (en) * | 2012-05-31 | 2012-08-09 | 국방과학연구소 | Method for detecting maritime targets |
| EP2905722A1 (en) * | 2014-02-10 | 2015-08-12 | Huawei Technologies Co., Ltd. | Method and apparatus for detecting salient region of image |
| CN104408974A (en) * | 2014-07-15 | 2015-03-11 | 王晓东 | Ship peripheral target detection alarm system |
| CN106161967A (en) * | 2016-09-13 | 2016-11-23 | 维沃移动通信有限公司 | A kind of backlight scene panorama shooting method and mobile terminal |
| CN106570487A (en) * | 2016-11-10 | 2017-04-19 | 维森软件技术（上海）有限公司 | Method and device for predicting collision between objects |
| JP6336693B1 (en) * | 2016-12-09 | 2018-06-06 | 株式会社日立国際電気 | Water intrusion detection system and method |
| WO2019011147A1 (en) * | 2017-07-10 | 2019-01-17 | Oppo广东移动通信有限公司 | Human face region processing method and apparatus in backlight scene |
| KR101800591B1 (en) * | 2017-10-30 | 2017-11-24 | 김용오 | System for vehicle number recognition using 3D camera |
| WO2019233393A1 (en) * | 2018-06-08 | 2019-12-12 | Oppo广东移动通信有限公司 | Image processing method and apparatus, storage medium, and electronic device |
| CN109242870A (en) * | 2018-07-13 | 2019-01-18 | 上海大学 | A kind of sea horizon detection method divided based on image with textural characteristics |
| CN109101897A (en) * | 2018-07-20 | 2018-12-28 | 中国科学院自动化研究所 | Object detection method, system and the relevant device of underwater robot |
| CN109978869A (en) * | 2019-03-29 | 2019-07-05 | 清华大学 | A kind of sea horizon detection method and system based on gray level co-occurrence matrixes and Hough transform |
| CN109993744A (en) * | 2019-04-09 | 2019-07-09 | 大连海事大学 | A kind of infrared target detection method under sea backlight environment |
| CN111489344A (en) * | 2020-04-10 | 2020-08-04 | 湖南索莱智能科技有限公司 | Method, system and related device for determining image definition |
| CN111898633A (en) * | 2020-06-19 | 2020-11-06 | 北京理工大学 | High-spectral image-based marine ship target detection method |
| CN112085753A (en) * | 2020-09-02 | 2020-12-15 | 广东海启星海洋科技有限公司 | Water level monitoring method, equipment, medium and monitoring system based on image processing |
Non-Patent Citations (1)
| Title |
|---|
| 一种基于海天线检测的舰船定位方法;黄英东;范宁军;李杰;;弹箭与制导学报;20081015(第05期);低86-288页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113705501A (en) | 2021-11-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113705501B (en) | Marine target detection method and system based on image recognition technology | |
| CN109086714A (en) | Table recognition method, identifying system and computer installation | |
| CN112819772A (en) | High-precision rapid pattern detection and identification method | |
| CN116110053B (en) | Container surface information detection method based on image recognition | |
| CN117197140B (en) | Irregular metal buckle forming detection method based on machine vision | |
| CN112070717B (en) | Power transmission line icing thickness detection method based on image processing | |
| CN110298796B (en) | Low-illumination image enhancement method based on improved Retinex and logarithmic image processing | |
| CN116152115B (en) | Garbage image denoising processing method based on computer vision | |
| CN115984134A (en) | Intelligent enhancing method for remote sensing mapping image | |
| CN113011433B (en) | Filtering parameter adjusting method and device | |
| Leng | An improved non-local means algorithm for image denoising | |
| CN109829858A (en) | A kind of shipborne radar image spilled oil monitoring method based on local auto-adaptive threshold value | |
| CN115861359A (en) | Self-adaptive segmentation and extraction method for water surface floating garbage image | |
| CN112001858A (en) | Method and system for processing cytopathology image | |
| CN116485686A (en) | Sewage treatment image enhancement method by activated sludge process | |
| Meenu et al. | Evolution of new integrated haze removal algorithm based on haze line | |
| CN112184619A (en) | Metal part surface defect detection method based on deep learning | |
| Samanta et al. | Statistical approach for Classification of SAR Images | |
| Song et al. | Design of Video Pre-processing Algorithm for High-speed Processing of Maritime Object Detection System and Deep Learning based Integrated System | |
| CN116188990A (en) | Unmanned aerial vehicle remote sensing-based earth surface vegetation identification method and system | |
| CN115841632A (en) | Power transmission line extraction method and device and binocular ranging method | |
| CN115908155A (en) | NSST domain combined GAN and scale correlation coefficient low-illumination image enhancement and denoising method | |
| Cui et al. | Single image haze removal based on luminance weight prior | |
| CN114998186A (en) | Image processing-based method and system for detecting surface scab defect of copper starting sheet | |
| Gautam et al. | A GUI for automatic extraction of signature from image document |
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 |