CN117557785A - Image processing-based long-distance fishing boat plate recognition method - Google Patents

Image processing-based long-distance fishing boat plate recognition method Download PDF

Info

Publication number
CN117557785A
CN117557785A CN202410038113.4A CN202410038113A CN117557785A CN 117557785 A CN117557785 A CN 117557785A CN 202410038113 A CN202410038113 A CN 202410038113A CN 117557785 A CN117557785 A CN 117557785A
Authority
CN
China
Prior art keywords
areas
individual
region
ship board
area
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.)
Granted
Application number
CN202410038113.4A
Other languages
Chinese (zh)
Other versions
CN117557785B (en
Inventor
叶宁
乐仁龙
楼杭欣
楼靖娟
陈丽巧
雷景生
杨胜英
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Ningbo Haihaixian Information Technology Co ltd
Original Assignee
Ningbo Haihaixian Information Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Ningbo Haihaixian Information Technology Co ltd filed Critical Ningbo Haihaixian Information Technology Co ltd
Priority to CN202410038113.4A priority Critical patent/CN117557785B/en
Publication of CN117557785A publication Critical patent/CN117557785A/en
Application granted granted Critical
Publication of CN117557785B publication Critical patent/CN117557785B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Evolutionary Computation (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the technical field of image processing, in particular to a long-distance fishing boat plate identification method based on image processing, which comprises the following steps: collecting a fishing boat image, and dividing the fishing boat image to obtain a plurality of areas; according to the plurality of areas, acquiring a first possibility parameter, a second possibility parameter and a third possibility parameter of all the areas being the ship board areas; acquiring the possibility parameters of all the areas being the ship board areas according to the first possibility parameters, the second possibility parameters and the third possibility parameters of all the areas being the ship board areas; acquiring the area to be enhanced according to the possibility parameters of all the areas being the ship board areas; reinforcing the region to be reinforced to obtain a reinforcing result of the region to be reinforced; and identifying the information on the ship board according to the enhancement result of the area to be enhanced. According to the invention, the ship board area is screened out from a large number of areas through the unique characteristics of the ship board area, and the information on the ship board can be identified by enhancing the areas independently.

Description

Image processing-based long-distance fishing boat plate recognition method
Technical Field
The invention relates to the technical field of image processing, in particular to a long-distance fishing boat plate recognition method based on image processing.
Background
Because when gathering fishing boat tablet image, can be because gathering the distance and lead to the tablet image to blur to the information on the discernment tablet, and in order to discern the information on the tablet accurately, need carry out image enhancement to the tablet, but traditional image enhancement technique is according to whole image to carry out the enhancement to the image, and the enhancement result of this moment is although for whole image more clear, but to the tablet can't obtain good enhancement effect, traditional image enhancement technique can't carry out effectual enhancement to the tablet promptly.
Disclosure of Invention
The invention provides a long-distance fishing boat plate identification method based on image processing, which aims to solve the existing problems: the conventional image enhancement technology cannot effectively enhance the ship board.
The invention relates to a long-distance fishing boat plate recognition method based on image processing, which adopts the following technical scheme:
the method comprises the following steps:
collecting a fishing boat image, and dividing the fishing boat image to obtain a plurality of areas;
acquiring the reference degree of each channel in the region according to the difference between the channels in the region; acquiring a first probability parameter of all areas being ship board areas according to the reference degree of each channel in the areas; acquiring second probability parameters of all areas as ship board areas according to the shapes of the areas; acquiring gradient values of pixel points in the region and edge pixel points by utilizing an edge detection algorithm; acquiring third possibility parameters of all areas as ship board areas according to gradient values of pixel points in the areas and edge pixel points;
acquiring the possibility parameters of all the areas being the ship board areas according to the first possibility parameters of all the areas being the ship board areas, the second possibility parameters of all the areas being the ship board areas and the third possibility parameters of all the areas being the ship board areas;
acquiring the area to be enhanced according to the possibility parameters of all the areas being the ship board areas; reinforcing the region to be reinforced to obtain a reinforcing result of the region to be reinforced; and identifying the information on the ship board according to the enhancement result of the area to be enhanced.
Preferably, the fishing boat image is collected and segmented to obtain a plurality of areas, and the specific method comprises the following steps:
the fishing boat image is acquired through the color camera to obtain the fishing boat image, and the fishing boat image is segmented by using a selective search algorithm to obtain a plurality of areas.
Preferably, the obtaining the reference degree of each channel in the area according to the difference between channels in the area includes the following specific methods:
for the firstThe first region is obtained->Channel values of the individual channels of all pixels in the individual region according to +.>Channel values of each channel of all pixel points in each region; calculate->The reference degree of all channels in each area is calculated by the following specific formula:
in the method, in the process of the invention,indicate->Personal area->Reference degree of each channel; />Indicate->The number of pixels in the individual regions; />Indicate->First->The pixel point is at the +.>Channel values under the individual channels; />Indicating the number of channels.
Preferably, the first probability parameter that all the areas are ship board areas is obtained according to the reference degree of each channel in the areas, and the specific method comprises the following steps:
for the firstA region of->After the reference level of all channels of the individual region, from +.>The channel with the highest reference degree is selected as the +.>Reference channel of individual region, will +.>All channels in the individual region which are not reference channels are denoted by +.>Non-reference channels of the individual regions; will be->Reference degree of reference channel of individual region, and +.>The difference between the reference degree averages of all non-reference channels in the individual regions, denoted +.>Characteristic values of the individual regions; for->The characteristic values of the individual regions are subjected to linear normalization, and the linear normalization is completed>Characteristic value of individual region as +.>The individual zones are the first likelihood parameters for the ship board zone.
Preferably, the second probability parameters of all the areas being the ship board areas are obtained according to the shapes of the areas, and the specific method comprises the following steps:
for calculation of the firstThe second possibility parameter of the area being the area of the ship board is first of all the rotationThe Carlo algorithm obtains->The smallest circumscribed rectangle of the individual regions; then count +.>The number of pixels in the minimum bounding rectangle of the individual regions +.>The number of pixels in the individual region; according to->The number of pixels in the minimum bounding rectangle of the individual regions +.>The number of pixels in the individual region, calculating +.>The second probability parameter that each area is a ship board area is specifically calculated as follows:
in the method, in the process of the invention,indicate->The individual areas are second likelihood parameters for the ship board area; />Indicate->The number of pixels in the individual region; />Indicate->The number of pixels in the minimum bounding rectangle of the individual regions; />Representing a linear normalization function.
Preferably, the gradient value of the pixel point in the region and the edge pixel point are obtained by using an edge detection algorithm; according to the gradient values of the pixel points in the region and the edge pixel points, obtaining third possibility parameters of all the regions as the ship board regions, wherein the method comprises the following specific steps:
for the firstThe first of all for the->Subjecting the individual regions to graying treatment to obtain +.>Gray-scale map of individual region, then using Canny edge detection algorithm, obtain +.>Gradient value of each pixel point in gray level map of each region +.>Edge pixel points in the individual regions; according to->Gradient value of each pixel point in gray level map of each region +.>Edge pixel point in each region, calculate +.>The third possibility parameter of each area is a ship board area, and the specific calculation process is as follows:
in the method, in the process of the invention,indicate->The individual areas are third likelihood parameters of the ship board area; />Indicate->The number of pixels in the individual regions; />Indicate->The>Gradient values of the individual pixels; />Indicate->The number of edge pixels in each region; />Indicate->The number of pixels in the individual region; />Representing a linear normalization function.
Preferably, the obtaining the probability parameters of all the areas being the playing card areas according to the first probability parameters of all the areas being the playing card areas, the second probability parameters of all the areas being the playing card areas and the third probability parameters of all the areas being the playing card areas includes the following specific calculation formulas:
in the method, in the process of the invention,indicate->The individual areas are probability parameters of the ship board area; />Indicate->The individual areas are first likelihood parameters for the ship board area; />Indicate->The individual areas are second likelihood parameters for the ship board area; />Indicate->The individual zones are third likelihood parameters for the ship board zone.
Preferably, the obtaining the area to be reinforced according to the probability parameters that all the areas are ship board areas comprises the following specific methods:
presetting a probability parameter thresholdFor->Individual regions, if->The probability parameter of each area being a ship board area is greater than or equal to +.>Will be->The individual regions are denoted as regions to be enhanced.
Preferably, the enhancing the region to be enhanced to obtain an enhancing result of the region to be enhanced includes the following specific steps:
for enhancement of the firstThe first pixel point of all the areas to be enhanced>The number of channels is acquired at->No. H under individual channels>Maximum channel value and minimum channel value of pixel points in the regions to be enhanced, respectively +.>And->According to->And (3) withReinforcing->The first pixel point of all the areas to be enhanced>The specific calculation formulas of the channels are as follows:
in the method, in the process of the invention,indicate->The first part of the region to be enhanced>The +.>Channel values after the individual channel enhancement;indicate->The first part of the region to be enhanced>The +.>Channel values for the individual channels; />Representing a downward rounding operation;
and obtaining channel values of all channels of all pixel points in all areas to be enhanced, and completing enhancement of all areas to be enhanced.
Preferably, the identifying the information on the ship board according to the enhancement result of the region to be enhanced comprises the following specific methods:
inputting all the reinforced areas to be reinforced into a neural network for character recognition, and using a Mask-RCNN network; executing the identification task, and adopting a loss function as follows: a cross entropy loss function; the inputs to the neural network are: all the reinforced areas to be reinforced; the output is: information on the ship board.
The technical scheme of the invention has the beneficial effects that: according to the invention, the fishing boat image is divided into a plurality of areas, and the characteristics of the boat-plate areas on the channel, the shape and the pixel gradient are combined to obtain the first possibility parameter, the second possibility parameter and the third possibility parameter of the boat-plate areas, so that the possibility parameter of the boat-plate areas is obtained, and the information on the boat-plate can be identified by screening the areas according to the possibility parameter of the boat-plate areas, so that the areas to be enhanced are obtained.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of steps of a method for identifying a long-distance fishing boat placard based on image processing;
fig. 2 is a schematic diagram of a fishing boat board according to an embodiment of the present invention.
Detailed Description
In order to further describe the technical means and effects adopted by the invention to achieve the preset aim, the following is a detailed description of a specific implementation, structure, characteristics and effects of the method for identifying a long-distance fishing boat plate based on image processing according to the invention with reference to the accompanying drawings and the preferred embodiment. In the following description, different "one embodiment" or "another embodiment" means that the embodiments are not necessarily the same. Furthermore, the particular features, structures, or characteristics of one or more embodiments may be combined in any suitable manner.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
The invention provides a specific scheme of a long-distance fishing boat plate recognition method based on image processing, which is specifically described below with reference to the accompanying drawings.
Referring to fig. 1, a flowchart of steps of a method for identifying a long-distance fishing boat board based on image processing according to an embodiment of the present invention is shown, the method includes the following steps:
step S001: and acquiring a fishing boat image, and dividing the fishing boat image to obtain a plurality of areas.
It should be noted that, as a remote ship board recognition method based on image processing, the final purpose of this embodiment is to accurately recognize information on a ship board, so that it is first necessary to collect a fishing ship image, then analyze the fishing ship image, and recognize information on the ship board.
Specifically, a fishing boat image is acquired through a color camera, and the fishing boat image is obtained.
It should be further noted that, as a remote ship board recognition method based on image processing, namely, the acquired fishing ship image is too far in acquisition distance, so that the ship board in the fishing ship image is blurred, and the information on the ship board is difficult to recognize, but in order to accurately recognize the information on the ship board, the traditional image enhancement technology needs to enhance the image according to the whole image, and the enhancement result is clearer for the whole image, but the ship board cannot be enhanced well, namely, the traditional image enhancement technology cannot enhance the ship board effectively; therefore, the embodiment provides an image enhancement method, firstly, according to a fishing boat image, a ship plate area in the fishing boat image is obtained, and then, the ship plate area is subjected to image enhancement independently, so that the enhancement effect of the ship plate is optimal; therefore, after the fishing boat image is obtained, it is necessary to divide the fishing boat image into regions so that the boat license region can be obtained.
Specifically, the fishing boat image is segmented by using a selective search algorithm, so as to obtain a plurality of areas.
It should be further noted that, since the selective search algorithm is a well-known prior art, the present embodiment will not be described in detail, and in several areas, there is one ship board area.
To this end, several areas are obtained.
Step S002: acquiring the reference degree of each channel in the region according to the difference between the channels in the region; acquiring a first probability parameter of all areas being ship board areas according to the reference degree of each channel in the areas; acquiring second probability parameters of all areas as ship board areas according to the shapes of the areas; acquiring gradient values of pixel points in the region and edge pixel points by utilizing an edge detection algorithm; and acquiring third possibility parameters of all the areas as the ship board areas according to the gradient values of the pixel points in the areas and the edge pixel points.
It should be noted that, because the present embodiment is used as a remote ship board identification method based on image processing, the ship board area is individually enhanced with an image, so that the enhancement effect of the ship board is optimal, and finally, the information on the ship board can be accurately identified. In the areas obtained in step S001, there are areas of the ship plate, and therefore, the ship plate areas need to be selected from the areas.
It should be further noted that, in the "regulations for ship name of fishery ships", the ship-board production of fishing ships requires that the color contrast between the ship-board background and the license plate is strong, as shown in fig. 2, so that the difference between the channels in the ship-board areas in several areas is larger than the difference between the channels in the non-ship-board areas; on the basis of this, a first probability parameter for all areas to be ship board areas can be obtained.
Specifically, for the firstThe first region is obtained->Channel values of the individual channels of all pixels in the individual region according to +.>Channel values of each channel of all pixel points in each region; calculate->The reference degree of all channels in each area is calculated by the following specific formula:
in the method, in the process of the invention,indicate->Personal area->Reference degree of each channel; />Indicate->The number of pixels in the individual regions; />Indicate->First->The pixel point is at the +.>Channel values under the individual channels; />Indicating the number of channels.
Note that, the channel in the present embodiment refers to a color channel in a color image, and the channel value refers to a color value in the channel.
Obtain the firstAfter the reference level of all channels of the individual region, from +.>The channel with the highest reference degree is selected as the +.>Reference channel of individual region, will +.>All channels in the individual region which are not reference channels are denoted by +.>Non-reference channels of the individual regions; will be->Reference degree of reference channel of individual region, and +.>The difference between the reference degree averages of all non-reference channels in the individual regions, denoted +.>Characteristic values of the individual regions; for->The characteristic values of the individual regions are subjected to linear normalization, and the linear normalization is completed>Characteristic value of individual region as +.>The first probability parameter of the area being the ship board area is marked as +.>。
It should be further noted that,the greater the value of +.>The higher the likelihood that an area is a ship board area.
So far, the first probability parameters that all the areas are the ship board areas are obtained.
It should be noted that, as a remote ship board identification method based on image processing, the present embodiment aims to accurately identify information on a ship board, and in order to more accurately identify information on a ship board, the present embodiment further obtains second probability parameters that all areas are ship board areas according to shape features of the ship board.
It should be further noted that, as shown in fig. 2, since the shape of the ship board is a rectangle, the ship board area has a certain rectangular feature in shape, and thus, the second probability parameter that all areas are ship board areas can be obtained based on this.
Specifically, for the calculation ofThe second possibility parameter of each area being a ship board area is obtained by first utilizing a rotary clamping algorithm>The specific process of obtaining the minimum circumscribed rectangle of each region is taken as a known prior art according to the rotating shell-clamping algorithm, so that the description of the embodiment is not repeated; then count +.>The number of pixels in the minimum bounding rectangle of the individual regions +.>The number of pixels in the individual region; according to->The number of pixels in the minimum bounding rectangle of the individual regions +.>The number of pixels in the individual region, calculating +.>The second probability parameter that each area is a ship board area is specifically calculated as follows:
in the method, in the process of the invention,indicate->The individual areas are second likelihood parameters for the ship board area; />Indicate->The number of pixels in the individual region; />Indicate->The number of pixels in the minimum bounding rectangle of the individual regions; />Representing a linear normalization function.
It should be further noted that,the greater the value of +.>The higher the likelihood that an area is a ship board area.
So far, the second probability parameters that all the areas are the ship board areas are obtained.
It should be noted that, the present embodiment is used as a remote ship board identification method based on image processing, which aims to accurately identify information on a ship board, and in order to more accurately identify information on a ship board, further, according to a gradient value of a pixel point in a ship board area, the present embodiment obtains third possibility parameters that all areas are ship board areas.
It should be further noted that, as shown in fig. 2, the color contrast between the tile background and the license plate is strong, so that there are a large number of pixels with large gradient values in the tile region, that is, the gradient values of the pixels in the tile region are large and the ratio of the edge pixels is high, so that the third probability parameter that all the regions are tile regions can be obtained based on the gradient values.
Specifically, for the calculation ofThe third possible parameter for the area being the ship board area is first of all for +.>Subjecting the individual regions to graying treatment to obtain +.>Gray-scale map of individual region, then using Canny edge detection algorithm, obtain +.>Gradient value of each pixel point in gray level map of each region +.>The gradient value of the pixel point and the specific process of the edge pixel point are obtained by using the Canny edge detection algorithm as a well-known prior art, so that redundant description is omitted in the embodiment; according to->Gradient value of each pixel point in gray level map of each region +.>Edge pixel point in each region, calculate +.>The third possibility parameter of each area is a ship board area, and the specific calculation process is as follows:
in the method, in the process of the invention,indicate->The individual areas are third likelihood parameters of the ship board area; />Indicate->The number of pixels in the individual regions; />Indicate->The>Gradient values of the individual pixels; />Indicate->The number of edge pixels in each region; />Indicate->The number of pixels in the individual region; />Representing a linear normalization function.
It should be further noted that,the greater the value of +.>The higher the likelihood that an area is a ship board area.
So far, a third probability parameter that all the areas are ship board areas is obtained.
Step S003: and acquiring the possibility parameters of all the areas being the ship board areas according to the first possibility parameters of all the areas being the ship board areas, the second possibility parameters of all the areas being the ship board areas and the third possibility parameters of all the areas being the ship board areas.
It should be noted that, because the present embodiment is used as a remote ship board identification method based on image processing, the ship board area is individually enhanced with an image, so that the enhancement effect of the ship board is optimal, and finally, the information on the ship board can be accurately identified. Step S002 is performed to obtain a first probability parameter that all areas are the playing card areas, a second probability parameter that all areas are the playing card areas, and a third probability parameter that all areas are the playing card areas; and obtaining the possibility parameters of all the areas being the ship board areas according to the first possibility parameters of all the areas being the ship board areas, the second possibility parameters of all the areas being the ship board areas and the third possibility parameters of all the areas being the ship board areas.
Specifically, for acquisition of the firstThe individual zones are probability parameters for a ship board zone,according to->The individual area is the first probability parameter, the +.sup.th of the ship board area>The second probability parameter of the area being the ship board area +.>The third possibility parameter with the individual area being the ship board area, the +.>The individual areas are probability parameters of the ship board areas, and a specific calculation formula is as follows:
in the method, in the process of the invention,indicate->The individual areas are probability parameters of the ship board area; />Indicate->The individual areas are first likelihood parameters for the ship board area; />Indicate->The individual areas are second likelihood parameters for the ship board area; />Indicate->The individual zones are third likelihood parameters for the ship board zone.
It should be further noted that since、/>And +.>The greater the value of +.>The higher the likelihood that an individual area is a ship board area; thus->The greater the value of +.>The higher the likelihood that an area is a ship board area.
So far, the probability parameters that all the areas are ship board areas are obtained.
Step S004: acquiring the area to be enhanced according to the possibility parameters of all the areas being the ship board areas; reinforcing the region to be reinforced to obtain a reinforcing result of the region to be reinforced; and identifying the information on the ship board according to the enhancement result of the area to be enhanced.
It should be noted that, since the present embodiment is used as a remote ship board recognition method based on image processing, the final purpose is to realize that information on a ship board can be accurately recognized. In this embodiment, the information on the ship board is accurately identified through the neural network, so that the information on the ship board can be accurately identified, and therefore, all areas are screened according to the possibility parameters that all areas are ship board areas.
Specifically, a probability parameter threshold is preset，/>The specific value of (2) can be set by combining with the actual situation, the hard requirement is not required in the embodiment, and +_ is adopted in the embodiment>To describe, for the->A plurality of regions; if->The probability parameter for the individual area being the ship board area is less than +.>Then->The individual areas must not be ship board areas; if->The probability parameter of each area being a ship board area is greater than or equal to +.>Then->The individual areas may be ship board areas; the area that may be a ship board is noted as the area to be enhanced.
It should be noted that, a threshold is preset to screen out the area which is not necessarily the ship board area, so as to avoid interfering with the judgment of the neural network. And after the areas to be enhanced are obtained, each area to be enhanced can be enhanced, and all the enhanced areas to be enhanced are input into the neural network, so that the purpose of accurately identifying the information on the ship board is achieved.
It should be further noted that, since there are several channels in the region to be enhanced, each channel needs to be enhanced when the region to be enhanced is enhanced, and since the process of enhancing each channel is identical, in this embodiment, enhancement of any channel in any region to be enhanced is described as an example.
In particular, for enhancement ofThe first pixel point of all the areas to be enhanced>The number of channels is acquired at->No. H under individual channels>Maximum channel value and minimum channel value of pixel points in the regions to be enhanced, respectively +.>And->According toAnd->Reinforcing->The first pixel point of all the areas to be enhanced>The specific calculation formulas of the channels are as follows:
in the method, in the process of the invention,indicate->The first part of the region to be enhanced>The +.>Channel values after the individual channel enhancement;indicate->The first part of the region to be enhanced>The +.>Channel values for the individual channels; />Is indicated at +.>No. H under individual channels>Maximum channel values of pixel points in the areas to be enhanced; />Is indicated at +.>No. H under individual channels>Minimum channel values of pixel points in the areas to be enhanced; />Representing a rounding down operation.
And similarly, obtaining channel values of all channels of all pixel points in all areas to be enhanced, and completing enhancement of all areas to be enhanced.
After the enhancement of all the areas to be enhanced is completed, all the enhanced areas to be enhanced can be input into the neural network, and information on the ship board is identified.
Specifically, all the enhanced regions to be enhanced are input into a neural network for character recognition, and a Mask-RCNN network is used. Executing the identification task, and adopting a loss function as follows: a cross entropy loss function; the inputs to the neural network are: all the reinforced areas to be reinforced; the output is: information on the ship board.
This embodiment is completed.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the invention, but any modifications, equivalent substitutions, improvements, etc. within the principles of the present invention should be included in the scope of the present invention.

Claims (10)

1. The method for identifying the long-distance fishing boat plate based on the image processing is characterized by comprising the following steps of:
collecting a fishing boat image, and dividing the fishing boat image to obtain a plurality of areas;
acquiring the reference degree of each channel in the region according to the difference between the channels in the region; acquiring a first probability parameter of all areas being ship board areas according to the reference degree of each channel in the areas; acquiring second probability parameters of all areas as ship board areas according to the shapes of the areas; acquiring gradient values of pixel points in the region and edge pixel points by utilizing an edge detection algorithm; acquiring third possibility parameters of all areas as ship board areas according to gradient values of pixel points in the areas and edge pixel points;
acquiring the possibility parameters of all the areas being the ship board areas according to the first possibility parameters of all the areas being the ship board areas, the second possibility parameters of all the areas being the ship board areas and the third possibility parameters of all the areas being the ship board areas;
acquiring the area to be enhanced according to the possibility parameters of all the areas being the ship board areas; reinforcing the region to be reinforced to obtain a reinforcing result of the region to be reinforced; and identifying the information on the ship board according to the enhancement result of the area to be enhanced.
2. The method for identifying the long-distance fishing boat cards based on image processing according to claim 1, wherein the steps of collecting the fishing boat images and dividing the fishing boat images to obtain a plurality of areas comprise the following specific steps:
the fishing boat image is acquired through the color camera to obtain the fishing boat image, and the fishing boat image is segmented by using a selective search algorithm to obtain a plurality of areas.
3. The method for identifying the long-distance fishing boat cards based on the image processing according to claim 1, wherein the method for obtaining the reference degree of each channel in the area according to the difference between the channels in the area comprises the following specific steps:
for the firstThe first region is obtained->Channel values of the individual channels of all pixels in the individual region according to +.>Channel values of each channel of all pixel points in each region; calculate->The reference degree of all channels in each area is calculated by the following specific formula:
in the method, in the process of the invention,indicate->Personal area->Reference degree of each channel; />Indicate->The number of pixels in the individual regions;indicate->First->The pixel point is at the +.>Channel values under the individual channels; />Indicating the number of channels.
4. The method for identifying the long-distance fishing boat nameplate based on the image processing as claimed in claim 3, wherein the method for obtaining the first probability parameter that all the areas are nameplate areas according to the reference degree of each channel in the areas comprises the following specific steps:
for the firstA region of->After the reference level of all channels of the individual region, from +.>The channel with the highest reference degree is selected as the +.>Reference channel of individual region, will +.>All channels in the individual region which are not reference channels are denoted by +.>Non-reference channels of the individual regions; will be->Reference degree of reference channel of individual region, and +.>The difference between the reference degree averages of all non-reference channels in the individual regions, denoted +.>Characteristic values of the individual regions; for->The characteristic values of the individual regions are subjected to linear normalization, and the linear normalization is completed>Characteristic value of individual region as +.>The individual zones are the first likelihood parameters for the ship board zone.
5. The method for identifying the long-distance fishing boat nameplate based on the image processing according to claim 1, wherein the method for obtaining the second probability parameter that all the areas are nameplate areas according to the shape of the areas comprises the following specific steps:
for calculation of the firstThe second possibility parameter of each area being a ship board area is obtained by first utilizing a rotary clamping algorithm>The smallest circumscribed rectangle of the individual regions; then count +.>The number of pixels in the minimum bounding rectangle of the individual regions +.>The number of pixels in the individual region; according to->The number of pixels in the minimum bounding rectangle of the individual regions +.>The number of pixels in the individual region, calculating +.>The second probability parameter that each area is a ship board area is specifically calculated as follows:
in the method, in the process of the invention,indicate->The individual areas are second likelihood parameters for the ship board area; />Indicate->The number of pixels in the individual region; />Indicate->The number of pixels in the minimum bounding rectangle of the individual regions; />Representing a linear normalization function.
6. The method for identifying the long-distance fishing boat plate based on the image processing according to claim 1, wherein the gradient value of the pixel points in the region and the edge pixel points are obtained by utilizing an edge detection algorithm; according to the gradient values of the pixel points in the region and the edge pixel points, obtaining third possibility parameters of all the regions as the ship board regions, wherein the method comprises the following specific steps:
for the firstThe first of all for the->Subjecting the individual regions to graying treatment to obtain +.>Gray-scale map of individual region, then using Canny edge detection algorithm, obtain +.>Gradient value of each pixel point in gray level map of each region +.>Edge pixel points in the individual regions; according to->Gradient value of each pixel point in gray level map of each region +.>Edge pixel point in each region, calculate +.>The third possibility parameter of each area is a ship board area, and the specific calculation process is as follows:
in the method, in the process of the invention,indicate->The individual areas are third likelihood parameters of the ship board area; />Indicate->The number of pixels in the individual regions; />Indicate->The>Gradient values of the individual pixels; />Indicate->The number of edge pixels in each region;indicate->The number of pixels in the individual region; />Representing a linear normalization function.
7. The method for identifying a long-distance fishing boat nameplate based on image processing according to claim 1, wherein the obtaining the probability parameters of all areas being nameplate areas according to the first probability parameters of all areas being nameplate areas, the second probability parameters of all areas being nameplate areas and the third probability parameters of all areas being nameplate areas comprises the following specific calculation formulas:
in the method, in the process of the invention,indicate->The individual areas are probability parameters of the ship board area; />Indicate->The individual areas are first likelihood parameters for the ship board area; />Indicate->The individual areas are second likelihood parameters for the ship board area; />Indicate->The individual zones are third likelihood parameters for the ship board zone.
8. The method for identifying the long-distance fishing boat billboards based on image processing according to claim 1, wherein the obtaining the area to be enhanced according to the probability parameters of all areas being the billboards area comprises the following specific steps:
presetting a probability parameter thresholdFor->Individual regions, if->The probability parameter of each area being a ship board area is greater than or equal to +.>Will be->The individual areas are denoted as areas to be enhancedDomain.
9. The method for identifying the long-distance fishing boat plate based on the image processing according to claim 1, wherein the method for enhancing the region to be enhanced to obtain the enhancement result of the region to be enhanced comprises the following specific steps:
for enhancement of the firstThe first pixel point of all the areas to be enhanced>The number of channels is acquired at->No. H under individual channels>Maximum channel value and minimum channel value of pixel points in the regions to be enhanced, respectively +.>And->According to->And (3) withReinforcing->The first pixel point of all the areas to be enhanced>The specific calculation formulas of the channels are as follows:
in the method, in the process of the invention,indicate->The first part of the region to be enhanced>The +.>Channel values after the individual channel enhancement; />Indicate->The first part of the region to be enhanced>The +.>Channel values for the individual channels; />Representing a downward rounding operation;
and obtaining channel values of all channels of all pixel points in all areas to be enhanced, and completing enhancement of all areas to be enhanced.
10. The method for identifying the long-distance fishing boat nameplate based on the image processing according to claim 1, wherein the identifying the information on the boat nameplate according to the enhancement result of the region to be enhanced comprises the following specific steps:
inputting all the reinforced areas to be reinforced into a neural network for character recognition, and using a Mask-RCNN network; executing the identification task, and adopting a loss function as follows: a cross entropy loss function; the inputs to the neural network are: all the reinforced areas to be reinforced; the output is: information on the ship board.
CN202410038113.4A 2024-01-11 2024-01-11 Image processing-based long-distance fishing boat plate recognition method Active CN117557785B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202410038113.4A CN117557785B (en) 2024-01-11 2024-01-11 Image processing-based long-distance fishing boat plate recognition method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202410038113.4A CN117557785B (en) 2024-01-11 2024-01-11 Image processing-based long-distance fishing boat plate recognition method

Publications (2)

Publication Number Publication Date
CN117557785A true CN117557785A (en) 2024-02-13
CN117557785B CN117557785B (en) 2024-04-02

Family

ID=89823564

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202410038113.4A Active CN117557785B (en) 2024-01-11 2024-01-11 Image processing-based long-distance fishing boat plate recognition method

Country Status (1)

Country Link
CN (1) CN117557785B (en)

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090306507A1 (en) * 2008-06-05 2009-12-10 Dong Gyu Hyun Anatomical Feature Extraction From An Ultrasound Liver Image
CN111899450A (en) * 2020-07-24 2020-11-06 宁波盛洋电子科技有限公司 Method and system for monitoring ships entering and exiting port and finding dangerous ships
CN111985475A (en) * 2020-08-26 2020-11-24 苏州工图智能科技有限公司 Ship board identification method, computing device and storage medium
CN113486895A (en) * 2021-07-16 2021-10-08 浙江华是科技股份有限公司 Ship board identification method and device and computer readable storage medium
CN113705505A (en) * 2021-09-02 2021-11-26 浙江索思科技有限公司 Marine fishery-oriented ship target detection method and system
CN114627034A (en) * 2020-12-08 2022-06-14 华为技术有限公司 Image enhancement method, training method of image enhancement model and related equipment
CN115797374A (en) * 2023-02-03 2023-03-14 长春理工大学 Airport runway extraction method based on image processing
WO2023065792A1 (en) * 2021-10-22 2023-04-27 杭州睿胜软件有限公司 Image processing method and apparatus, electronic device, and computer-readable storage medium
CN116704177A (en) * 2023-08-01 2023-09-05 东莞市零越传动科技有限公司 Gear box fault detection method based on image data
CN116935332A (en) * 2023-03-30 2023-10-24 华南理工大学 Fishing boat target detection and tracking method based on dynamic video
CN117058257A (en) * 2023-08-15 2023-11-14 中国传媒大学 Method and equipment for classifying and compressing region of interest of ship

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090306507A1 (en) * 2008-06-05 2009-12-10 Dong Gyu Hyun Anatomical Feature Extraction From An Ultrasound Liver Image
CN111899450A (en) * 2020-07-24 2020-11-06 宁波盛洋电子科技有限公司 Method and system for monitoring ships entering and exiting port and finding dangerous ships
CN111985475A (en) * 2020-08-26 2020-11-24 苏州工图智能科技有限公司 Ship board identification method, computing device and storage medium
CN114627034A (en) * 2020-12-08 2022-06-14 华为技术有限公司 Image enhancement method, training method of image enhancement model and related equipment
CN113486895A (en) * 2021-07-16 2021-10-08 浙江华是科技股份有限公司 Ship board identification method and device and computer readable storage medium
CN113705505A (en) * 2021-09-02 2021-11-26 浙江索思科技有限公司 Marine fishery-oriented ship target detection method and system
WO2023065792A1 (en) * 2021-10-22 2023-04-27 杭州睿胜软件有限公司 Image processing method and apparatus, electronic device, and computer-readable storage medium
CN115797374A (en) * 2023-02-03 2023-03-14 长春理工大学 Airport runway extraction method based on image processing
CN116935332A (en) * 2023-03-30 2023-10-24 华南理工大学 Fishing boat target detection and tracking method based on dynamic video
CN116704177A (en) * 2023-08-01 2023-09-05 东莞市零越传动科技有限公司 Gear box fault detection method based on image data
CN117058257A (en) * 2023-08-15 2023-11-14 中国传媒大学 Method and equipment for classifying and compressing region of interest of ship

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
JING LI ET AL.: "Enhancement of Target Feature Regions and Intention-Driven Visual Attention Selection in Traffic Scenes", 《2022 IEEE INTELLIGENT VEHICLES SYMPOSIUM (IV)》, 19 July 2022 (2022-07-19), pages 404 - 410 *
SHENGYING YANG ET AL.: "Trident-LK Net: A Lightweight Trident Structure Network With Large Kernel for Muti-Scale Defect Detection", 《IEEE ACCESS 》, vol. 11, 16 November 2023 (2023-11-16), pages 131073 *
吴书楷等: "结合深度特征迁移与融合的两阶段船牌定位算法", 《计算机辅助设计与图形学学报》, vol. 32, no. 04, 17 February 2020 (2020-02-17), pages 628 - 634 *
张晶晶: "细粒度车辆图像分析算法的研究与应用", 《中国优秀硕士学位论文全文数据库工程科技Ⅱ辑》, no. 06, 15 June 2022 (2022-06-15), pages 034 - 300 *
李丹等: "基于模糊理论的SAR图像海面油膜识别", 《海洋湖沼通报》, no. 02, 15 June 2010 (2010-06-15), pages 147 - 154 *
雷景生等: "改进YOLOX的轻量级多方向车牌检测算法", 《计算机工程与应用》, 30 November 2023 (2023-11-30), pages 1 - 14 *

Also Published As

Publication number Publication date
CN117557785B (en) 2024-04-02

Similar Documents

Publication Publication Date Title
CN108229526B (en) Network training method, network training device, image processing method, image processing device, storage medium and electronic equipment
CN107507173B (en) No-reference definition evaluation method and system for full-slice image
CN107038416B (en) Pedestrian detection method based on binary image improved HOG characteristics
CN114926436B (en) Defect detection method for periodic pattern fabric
CN116664559B (en) Machine vision-based memory bank damage rapid detection method
CN115049657A (en) Glass defect detection method
CN114820625B (en) Automobile top block defect detection method
CN115082465B (en) Wool and cashmere classification method based on scanning electron microscope image processing
CN110687122A (en) Method and system for detecting surface cracks of ceramic tile
CN117635609B (en) Visual inspection method for production quality of plastic products
CN117994154B (en) Intelligent image denoising method based on sensor
CN110807763A (en) Method and system for detecting ceramic tile surface bulge
CN116309649A (en) Throat lesion abnormal region segmentation system based on CT image
CN111079626B (en) Living body fingerprint identification method, electronic equipment and computer readable storage medium
CN104268845A (en) Self-adaptive double local reinforcement method of extreme-value temperature difference short wave infrared image
CN117765287A (en) Image target extraction method combining LWR and density clustering
CN117557785B (en) Image processing-based long-distance fishing boat plate recognition method
CN108009480A (en) A kind of image human body behavioral value method of feature based identification
CN104102911A (en) Image processing for AOI (automated optical inspection)-based bullet appearance defect detection system
CN112532938B (en) Video monitoring system based on big data technology
CN112435186B (en) Fingerprint image enhancement method based on double-rule matrix direction field
CN112085683B (en) Depth map credibility detection method in saliency detection
CN118229574B (en) Digital enhancement method for water supply and drainage design drawing
CN110647843A (en) Face image processing method
CN117314899B (en) Carbon fiber plate quality detection method based on image characteristics

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