WO2021189782A1 - 图像处理方法、系统、自动行走设备及可读存储介质 - Google Patents

图像处理方法、系统、自动行走设备及可读存储介质 Download PDF

Info

Publication number
WO2021189782A1
WO2021189782A1 PCT/CN2020/115847 CN2020115847W WO2021189782A1 WO 2021189782 A1 WO2021189782 A1 WO 2021189782A1 CN 2020115847 W CN2020115847 W CN 2020115847W WO 2021189782 A1 WO2021189782 A1 WO 2021189782A1
Authority
WO
WIPO (PCT)
Prior art keywords
pixel
initial
value
image
feature value
Prior art date
Application number
PCT/CN2020/115847
Other languages
English (en)
French (fr)
Inventor
朱绍明
任雪
Original Assignee
苏州科瓴精密机械科技有限公司
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 苏州科瓴精密机械科技有限公司 filed Critical 苏州科瓴精密机械科技有限公司
Publication of WO2021189782A1 publication Critical patent/WO2021189782A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/90—Dynamic range modification of images or parts thereof
    • G06T5/92—Dynamic range modification of images or parts thereof based on global image properties
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/90—Dynamic range modification of images or parts thereof
    • 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
    • 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
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20172—Image enhancement details
    • G06T2207/20192—Edge enhancement; Edge preservation

Definitions

  • the present invention relates to the technical field of image processing, and in particular to an image processing method, system, automatic walking device, and readable storage medium that can eliminate the shadow boundary of an image.
  • Self-propelled equipment such as lawn mowers, vacuum cleaners
  • recognize the boundaries through images but due to the light conditions, trees and houses will form varying degrees of shadow boundaries on the grass.
  • the shadow boundaries are interference information for image recognition, and automatic walking equipment is easy
  • the shadow boundary line is misjudged as a boundary, so that it is impossible to enter or exit the shadow area to complete the job.
  • shading can be achieved through image enhancement, such as homomorphic filtering, histogram equalization, image enhancement based on Laplacian, image enhancement based on logarithmic transformation, and image enhancement based on gamma transformation.
  • image enhancement such as homomorphic filtering, histogram equalization, image enhancement based on Laplacian, image enhancement based on logarithmic transformation, and image enhancement based on gamma transformation.
  • Chinese patent application CN201310689070.8 discloses an all-weather traffic image enhancement method based on brightness reference drift, which fully considers the relationship between the monitoring image and the light intensity and the shooting time, analyzes the overall and real-time changes of illumination, and obtains the brightness respectively. The reference curve and brightness are fed back in real time, and weighted to obtain the current brightness reference value. Before enhancing the traffic monitoring image, first convert the image from the RGB color space to the HSV color space.
  • the threshold for dividing the brightness area in the Chinese patent application CN201310689070.8 is the brightness reference value obtained by the brightness reference curve L(t), and the algorithm calculation amount of the brightness reference curve L(t) is large, which affects the image processing speed.
  • the present invention provides an image processing method, system, automatic walking equipment and readable storage medium, which avoids misjudgment of the boundary system caused by the shadow boundary of the image.
  • the present invention provides an image processing method; it includes the following steps:
  • Preprocess the original image to identify each pixel as a bright area pixel or a dark area pixel, and extract the initial H channel feature value, initial S channel feature value, and initial V channel feature value of each pixel;
  • the initial H channel feature value the initial S channel feature value, and the compensated V channel feature value, synthesis processing and conversion processing are performed to obtain an image with the shadow boundary removed.
  • the process of preprocessing the original image includes the following steps:
  • the method further includes:
  • the method further includes:
  • the bright area pixel used to calculate the average pixel intensity of the bright area is the bright area pixel point of the first pixel point combination;
  • the dark area pixel used to calculate the dark area average pixel brightness intensity is the second pixel point The combined dark area pixels.
  • a number of brightness intensity difference value segment intervals are preset, and the process of obtaining the brightness compensation value includes:
  • the initial V channel feature value of the bright area pixel and/or the dark area pixel is compensated, and the initial V channel feature value of the bright area pixel is weakened and compensated according to the brightness compensation value:
  • the characteristic value of the V channel after compensation the characteristic value of the initial V channel-the brightness compensation value.
  • the synthesis process is to obtain an HSV image with a shadowed boundary based on the H channel feature value, the S channel feature value, and the compensated V channel feature value; the conversion process is to convert the shadowed HSV image into the shadowed HSV image RGB image of the border.
  • the present invention also provides an image processing system, including:
  • the preprocessing module is used to preprocess the original image to identify each pixel as a pixel in a bright area or a pixel in a dark area, and to extract the initial H channel feature value, initial S channel feature value and Initial V channel characteristic value;
  • Brightness compensation module which is used to calculate the average pixel brightness intensity ValueB in the bright area according to the V channel characteristic value of the pixels in the bright area, and calculate the average pixel brightness intensity ValueS in the dark area according to the V channel characteristic value of the pixels in the dark area; and according to the bright area average Pixel brightness intensity and average pixel brightness intensity in dark areas obtain the brightness compensation value OffsetValue; perform compensation processing on the initial V channel feature value according to the brightness compensation value to obtain the compensated V channel feature value;
  • the image generation module is used to perform synthesis processing and conversion processing according to the initial H channel feature value, the initial S channel feature value, and the compensated V channel feature value to obtain an image with the shadow boundary removed.
  • the present invention also provides an autonomous walking device, which includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the image processing method when the computer program is executed by the processor.
  • the present invention also provides a readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the image processing method are implemented.
  • the present invention adjusts the characteristic value of the V channel by the brightness compensation value to eliminate the shadow boundary of the image, and avoid the misjudgment of the boundary system caused by the shadow boundary of the image.
  • the present invention can compare the brightness intensity value of the pixel in the binarized image with a preset threshold, and can identify bright area pixels/dark area pixels, and the value range of the preset threshold value is not limited.
  • the calculation amount of the preset algorithm set in the present invention is small, and the image processing speed is fast.
  • different brightness compensation values are obtained through the identification of the brightness intensity difference and the brightness intensity difference segmentation interval by setting the brightness intensity difference value segmentation interval, thereby realizing the adaptive compensation effect.
  • Fig. 1 is a flowchart of a first embodiment of an image processing method according to the present invention
  • FIG. 2 is a detailed flowchart of step S10 in FIG. 1;
  • Fig. 3 is a detailed flowchart of step S30 in Fig. 1.
  • FIG. 4 is a detailed flowchart of step S50 in FIG. 1;
  • Fig. 5 is a flowchart of a second embodiment of an image processing method according to the present invention.
  • Fig. 6 is a flowchart of a third embodiment of the image processing method of the present invention.
  • FIG. 7 is a block diagram of the image processing system of the present invention.
  • Figure 8 is the initial V channel image obtained after the original image is preprocessed
  • Figure 9 is a binarized image obtained after preprocessing Figure 8.
  • Figure 10 is an image of the shadowed boundary obtained after the compensation process.
  • the present invention provides an image processing method; it includes the following steps:
  • Step S10 Preprocess the original image to identify each pixel as a bright area pixel or a dark area pixel, and extract the initial H channel feature value, initial S channel feature value, and initial V channel feature value of each pixel ;
  • Step S20 Calculate the average pixel brightness intensity ValueB of the bright area according to the initial V channel feature value of the pixel in the bright area, and calculate the average pixel brightness value ValueS of the dark area according to the initial V channel feature value of the pixel in the dark area;
  • Step S30 Obtain the brightness compensation value OffsetValue according to the average pixel brightness intensity in the bright area and the average pixel brightness intensity in the dark area;
  • Step S40 Perform compensation processing on the initial V channel characteristic value according to the brightness compensation value to obtain the compensated V channel characteristic value
  • Step S50 Perform synthesis processing and conversion processing according to the initial H channel feature value, the initial S channel feature value, and the compensated V channel feature value to obtain an image with the shadow boundary removed.
  • the preprocessing in the step S10 includes binarization processing, the original image is binarized to obtain a binarized image, and the pixel points in the binarized image
  • the brightness intensity value is compared with the preset threshold value, and the pixel points in the bright area/the pixel points in the dark area are identified according to the comparison result.
  • the brightness intensity value of the pixel in the binarized image is either 0 or 255, and the value range of the preset threshold used to identify the pixel in the bright area/the pixel in the dark area is not limited, 0-255 (including 0, including 255) can be any value.
  • the comparison relationship can be set according to needs, for example, determining whether the brightness intensity value of the pixel is greater than a preset threshold, determining whether the brightness intensity value of the pixel is less than a preset threshold, and determining whether the brightness intensity value of the pixel is equal to the preset threshold.
  • step S10 further includes the following steps:
  • Step S110 Obtain an RGB image
  • Step S120 Convert the RGB image into an HSV image, and extract the feature value of each pixel in the HSV image, where the feature value includes an initial H channel feature value, an initial S channel feature value, and an initial V channel feature value;
  • Step S130 Obtain an initial V channel image according to the initial V channel feature value of each pixel in the HSV image (as shown in FIG. 8);
  • Step S140 Binarize the initial V channel image to obtain a binarized image (as shown in Fig. 9).
  • the initial V channel can be converted by the OTSU algorithm (that is, the Otsu algorithm, also known as the maximum inter-class difference method).
  • Image binarization processing
  • Step S150 Determine whether the brightness intensity value of the pixel in the binarized image is greater than the preset threshold; if the brightness intensity value of the pixel in the binarized image is greater than the preset threshold, perform step S160; if the pixel in the binarized image is If the brightness intensity value of the point is not greater than the preset threshold value, step S170 is executed;
  • the brightness intensity value of the pixel in the binarized image is either 0 or 255
  • the value range of the preset threshold used to identify the bright area pixel/dark area pixel is not limited, 0-255 (Including 0, but not including 255) any value is acceptable.
  • Two-dimensional coordinates are defined in the binary image, and each pixel has a unique coordinate value. Traverse the binarized image with coordinate values in a predetermined sequence until all pixels in the binarized image are judged to be completed.
  • Step S160 Identify the pixel point as a pixel point in the bright area, and obtain the initial V channel feature value of the initial V channel image according to the coordinate value of the pixel point;
  • Step S170 Identify the pixel point as a dark area pixel point, and obtain an initial V channel feature value of the initial V channel image according to the coordinate value of the pixel point.
  • a number of brightness intensity difference value segment intervals are preset, and different brightness compensation values are obtained by identifying the brightness intensity difference value and the brightness intensity difference value segment interval, thereby realizing adaptive compensation Effect.
  • the different brightness compensation values can be obtained by setting a number of preset brightness compensation values (fixed values), or can be obtained according to a preset algorithm according to the difference in brightness intensity.
  • a plurality of brightness intensity difference segment intervals are preset, and the step S30 further includes the following steps:
  • Step S320 Identify the brightness intensity difference segment interval corresponding to the brightness intensity difference DiffValue, and select the corresponding adjustment parameter K according to the brightness intensity difference segment interval;
  • the preset algorithm sets one adjustment parameter K, and can also set multiple adjustment parameters K1 and K2 as needed.
  • the adjustment parameter K1 is associated with DiffValue
  • the adjustment parameter K2 is The fixed value parameter is adjusted through the combination of multiple adjustment parameters and DiffValue, so as to obtain a better compensation effect.
  • the brightness intensity difference segment interval can be set as needed, and the segment compensation is performed through the brightness intensity difference segment interval.
  • the corresponding adjustment parameter K is automatically selected according to the size of the DiffValue, which has an adaptive compensation effect.
  • the value of the adjustment parameter K is not limited to the following data.
  • the number of brightness intensity difference segmentation intervals and the endpoint value of the interval can be set according to needs. For example, Table 1 sets four brightness intensity difference segmentation intervals with valueA, valueB, valueC, and valueD as the endpoints.
  • the step S40 performs compensation processing on the initial V channel characteristic values of the pixels in the bright area and/or the pixels in the dark area according to the brightness compensation value. For example, only the pixels in the bright area may be compensated. Compensate the initial V channel feature value of the pixel, or only the initial V channel feature value of the pixel in the dark area, or, at the same time, the initial V channel feature value of the pixel in the bright area and the pixel in the dark area. The initial V channel eigenvalues are compensated separately.
  • V channel characteristic value after compensation initial V channel characteristic value-brightness compensation value.
  • V channel feature value after compensation initial V channel feature value + brightness compensation value.
  • the step S50 further includes:
  • Step S510 Perform synthesis processing according to the characteristic value of the H channel, the characteristic value of the S channel, and the characteristic value of the compensated V channel to obtain an HSV image with the shadow boundary removed (as shown in FIG. 10);
  • Step S520 Perform conversion processing on the de-shaded HSV image to obtain an RGB image with the de-shaded boundary.
  • step S10 between the step S10 and the step S20, it further includes:
  • Step S15 Determine whether all the pixels are pixels in the bright area or in the dark area; if yes, end this process; if not, execute step S20.
  • step S15 The situations that need compensation and those that do not need compensation are screened through step S15, so as to optimize the compensation process.
  • the present invention provides an image processing method; it includes the following steps:
  • Step S10 Preprocess the original image to identify each pixel as a bright area pixel or a dark area pixel, and extract the initial H channel feature value, initial S channel feature value, and initial V channel feature value of each pixel ;
  • Step S16 Filter the preprocessed pixels according to the S channel feature value, where all the bright area pixels in the preprocessed pixels are the first pixel point combination, and all the dark area pixels in the preprocessed pixel points are the second Pixel combination;
  • Step S20A Calculate the average brightness intensity ValueB of bright area pixels according to the initial V channel characteristic values of the bright area pixels of the first pixel point combination, and calculate the dark area average according to the initial V channel characteristic values of the dark area pixels of the second pixel point combination Pixel brightness intensity ValueS;
  • Step S30 Obtain the brightness compensation value OffsetValue according to the average pixel brightness intensity in the bright area and the average pixel brightness intensity in the dark area;
  • Step S40 Perform compensation processing on the initial V channel characteristic value according to the brightness compensation value to obtain the compensated V channel characteristic value
  • Step S50A Perform synthesis and conversion processing according to the initial H channel feature value, the initial S channel feature value, the initial V channel feature value, and the compensated V channel feature value to obtain an image with the shadow boundary removed.
  • Step S16 is used to filter and process the V channel feature value in part of the image from the compensation process, that is, before and after the compensation process, part of the image still retains the original V channel Eigenvalues. Through selective compensation processing, to avoid the loss of information in part of the image.
  • part of the original image is the car body
  • the present invention also provides an image processing system 1, including:
  • the preprocessing module 10 is used to preprocess the original image to identify each pixel as a bright area pixel or a dark area pixel, and to extract the initial H channel feature value and initial S channel feature value of each pixel And the initial V channel characteristic value;
  • the brightness compensation module 20 is used to calculate the average pixel brightness intensity ValueB of the bright area according to the V channel characteristic value of the pixel point in the bright area, and calculate the average pixel brightness intensity ValueS of the dark area according to the V channel characteristic value of the pixel point in the dark area; and according to the bright area
  • the average pixel brightness intensity and the average pixel brightness intensity in the dark area obtain the brightness compensation value OffsetValue; the initial V channel feature value is compensated according to the brightness compensation value to obtain the compensated V channel feature value;
  • the image generation module 30 is configured to perform synthesis processing and conversion processing according to the initial H channel feature value, the initial S channel feature value, and the compensated V channel feature value to obtain an image with the shadow boundary removed.
  • the present invention also provides an autonomous walking device, including a memory and a processor, the memory stores a computer program, and the processor implements the steps of the image processing method when the computer program is executed by the processor.
  • the automatic walking equipment of the present invention can be an automatic lawn mower, or an automatic vacuum cleaner, etc., or other equipment, such as spraying equipment, snow removal equipment, monitoring equipment, etc., suitable for unattended equipment.
  • the present invention also provides a readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the image processing method are implemented.
  • the present invention adjusts the characteristic value of the V channel through the brightness compensation value to eliminate the shadow boundary of the image, and avoid the boundary system misjudgment caused by the shadow boundary of the image.
  • the present invention can compare the brightness intensity value of the pixel in the binarized image with a preset threshold, and can identify bright area pixels/dark area pixels, and the value range of the preset threshold value is not limited.
  • the calculation amount of the preset algorithm set in the present invention is small, and the image processing speed is fast.
  • different brightness compensation values are obtained through the identification of the brightness intensity difference and the brightness intensity difference segmentation interval by setting the brightness intensity difference value segmentation interval, thereby realizing the adaptive compensation effect.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Multimedia (AREA)
  • Image Processing (AREA)

Abstract

一种图像处理方法、系统、自动行走设备及可读存储介质，所述图像处理方法包括以下步骤：将原始图像进行预处理以识别每个像素点为亮区像素点或暗区像素点，并提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值(S10)；计算亮区平均像素亮度强度ValueB与暗区平均像素亮度强度ValueS(S20)；根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue(S30)；根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值(S40)；根据初始H通道特征值、初始S通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像(S50)。通过亮度补偿值调整V通道特征值以消除图像的阴影边界，避免因图像的阴影边界导致边界系统误判。

Description

图像处理方法、系统、自动行走设备及可读存储介质 技术领域
本发明涉及图像处理技术领域，尤其涉及一种可消除图像的阴影边界的图像处理方法、系统、自动行走设备及可读存储介质。
背景技术
自动行走设备(例如割草机，吸尘器)通过图像识别边界，但由于光线条件使树、房屋会在草地上形成不同程度的阴影分界线，阴影分界线属于图像识别的干扰信息，自动行走设备容易将阴影分界线误判为边界，从而无法进入或者走出阴影区域完成作业。
去阴影在一定程度上可以通过图像增强实现，如同态滤波、直方图均衡化、基于拉普拉斯算子的图像增强、基于对数变换的图像增强以及基于伽马变换的图像增强。例如，中国专利申请CN201310689070.8揭露一种基于亮度基准漂移的全天候交通图像增强方法，充分考虑监控图像与光照强度和拍摄时间之间的相互关系，分析光照的总体变化和实时变化，分别得到亮度基准曲线和亮度实时反馈，并加权得到当前时刻的亮度基准值，在对交通监控图像增强之前，先把图像从RGB色彩空间转换到HSV色彩空间，在保持图像色度信息不变的基础上，运用亮度基准值对亮度分量分割，得到低亮度区和高亮度区，并分别求取每个亮度级的漂移参数，每个亮度级乘以对应亮度级的漂移参数得到增强后的亮度级，最后图像转换到RGB色彩空间得到增强后的图片。
但是，中国专利申请CN201310689070.8中用于亮度区域划分的阈值为通过亮度基准曲线L(t)取得的亮度基准值，亮度基准曲线L(t)的算法计算量大，影响图片处理速度。
发明内容
本发明提供一种图像处理方法、系统、自动行走设备及可读存储介质，避免因图像的阴影边界导致边界系统误判。
本发明提供一种图像处理方法；其包括以下步骤：
将原始图像进行预处理以识别每个像素点为亮区像素点或暗区像素点，并提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值；
根据亮区像素点的初始V通道特征值计算亮区平均像素亮度强度ValueB，根据暗区像素点的初始V通道特征值计算暗区平均像素亮度强度ValueS；
根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue；；
根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值；
根据初始H通道特征值、初始S通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像。
可选地，所述将原始图像进行预处理的过程包括以下步骤：
获取RGB图像；
将RGB图像转换为HSV图像，提取HSV图像中每个像素点的特征值，所述特征值包括初始H通道特征值、初始S通道特征值与初始V通道特征值；
根据HSV图像中每个像素点的初始V通道特征值以获得初始V通道图像；
将初始V通道图像进行二值化处理，获得二值化图像；
判断二值化图像中像素点的亮度强度值是否大于预设阈值，以识别每个像素点为亮区像素点或暗区像素点，二值化图像中像素点的亮度强度值大于预设阈值，则为亮区像素点，而二值化图像中像素点的亮度强度值不大于预设阈值，则为暗区像素点。
可选地，在识别亮区像素点或暗区像素点后，还包括：
判断所有像素点是否全为亮区像素点或暗区像素点；若是，则结束本次流程；若否，则继续执行后面的补偿流程。
可选地，在识别亮区像素点或暗区像素点后，还包括：
根据S通道特征值筛选出预处理的像素点，其中，预处理的像素点中所有亮区像素点为第一像素点组合，预处理的像素点中所有暗区像素点为第二像素点组合；
则后续流程中，用于计算亮区平均像素亮度强度的亮区像素点为第一像素点组合的亮区像素点；用于计算暗区平均像素亮度强度的暗区像素点为第二像素点组合的暗区像素点。
可选地，预设若干亮度强度差值分段区间，且所述获取亮度补偿值的过程包括：
计算亮区平均像素亮度强度ValueB与暗区平均像素亮度强度ValueS的亮度强度差值DiffValue＝ValueB-ValueS；
识别亮度强度差值DiffValue所对应的亮度强度差值分段区间，并根据亮度强度差值分段区间选取相应的调整参数K；
根据所述亮度强度差值DiffValue、调整参数K及预设算法获取亮度补偿值OffsetValue，且所述预设算法为：OffsetValue＝K*DiffValue。
可选地，根据亮度补偿值对亮区像素点和/或暗区像素点的初始V通道特征值进行补偿处理，根据亮度补偿值对亮区像素点的初始V通道特征值进行减弱补偿处理：补偿后V通道特征值＝初始V通道特征值-亮度补偿值，根据亮度补偿值对暗区像素点的初始V通道特征值进行增强补偿处理：补偿后V通道特征值＝初始V通道特征值+亮度补偿值。
可选地，所述合成处理为根据H通道特征值、S通道特征值、补偿后V通道特征值，以获得去阴影边界的HSV图像；所述转换处理为将去阴影HSV图像转换为去阴影边界的RGB图像。
本发明还提供一种图像处理系统，包括：
预处理模块，其用于将原始图像进行预处理以识别每个像素点为亮区像素点或暗区像素点，并用于提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值；
亮度补偿模块，其用于根据亮区像素点的V通道特征值计算亮区平均像素亮度强度ValueB，根据暗区像素点的V通道特征值计算暗区平均像素亮度强度ValueS；并根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue；根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值；
图像生成模块，其用于根据初始H通道特征值、初始S通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像。
本发明又提供一种自动行走设备，包括存储器和处理器，所述存储器存储 有计算机程序，且所述处理器执行所述计算机程序时实现所述图像处理方法的步骤。
本发明又提供一种可读存储介质，其上存储有计算机程序，且所述计算机程序被处理器执行时实现所述图像处理方法的步骤。
相较于现有技术，本发明通过亮度补偿值调整V通道特征值以消除图像的阴影边界，避免因图像的阴影边界导致边界系统误判。本发明可将二值化图像中的像素点的亮度强度值与预设阈值进行比较处理，可识别亮区像素点/暗区像素点，且所述预设阈值的取值范围不受限制。本发明中设置的预设算法的计算量小，图片处理速度快。本发明通过设置亮度强度差值分段区间，通过亮度强度差值与亮度强度差值分段区间的识别获取不同的亮度补偿值，从而实现自适应的补偿效果。
附图说明
图1为本发明图像处理方法的第一种实施例的流程图；
图2为图1中步骤S10的详细流程图；
图3为图1中步骤S30的详细流程图。
图4为图1中步骤S50的详细流程图；
图5为本发明图像处理方法的第二种实施例的流程图；
图6为本发明图像处理方法的第三种实施例的流程图；
图7为本发明图像处理系统的方框图；
图8为原始图像经预处理后获得的初始V通道图像；
图9为将图8进行预处理后获得的二值化图像；
图10为经补偿处理后获得的去阴影边界的图像。
具体实施方式
为了使本技术领域的人员更好地理解本发明中的技术方案，下面将结合本发明实施例中的附图，对本发明实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本发明一部分实施例，而不是全部的实施例。基于本发明中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都应当属于本发明保护的范围。
请参阅图1所示，本发明提供一种图像处理方法；其包括以下步骤：
步骤S10：将原始图像进行预处理以识别每个像素点为亮区像素点或暗区像素点，并提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值；
步骤S20：根据亮区像素点的初始V通道特征值计算亮区平均像素亮度强度ValueB，根据暗区像素点的初始V通道特征值计算暗区平均像素亮度强度ValueS；
步骤S30：根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue；
步骤S40：根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值；
步骤S50：根据初始H通道特征值、初始S通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像。
在本实用新型的另一个实施例中，所述步骤S10中的预处理包括二值化处 理，将原始图像经二值化处理后获得二值化图像，并将二值化图像中像素点的亮度强度值与预设阈值进行比较处理，根据比较结果以识别亮区像素点/暗区像素点。所述二值化图像中像素点的亮度强度值或为0，或为255，用于识别亮区像素点/暗区像素点的预设阈值的取值范围不受限制，0-255(包括0，包括255)的任意数值均可。可根据需要设置比较关系，例如，判断像素点的亮度强度值是否大于预设阈值，判断像素点的亮度强度值是否小于预设阈值，判断像素点的亮度强度值是否等于预设阈值。
请参阅图2所示，在本实用新型的另一个实施例中，所述步骤S10进一步包括以下步骤：
步骤S110：获取RGB图像；
步骤S120：将RGB图像转换为HSV图像，提取HSV图像中每个像素点的特征值，所述特征值包括初始H通道特征值、初始S通道特征值与初始V通道特征值；
步骤S130：根据HSV图像中每个像素点的初始V通道特征值以获得初始V通道图像(如图8所示)；
步骤S140：将初始V通道图像进行二值化处理，获得二值化图像(如图9所示)，例如，可通过OTSU算法(即大津算法，又称最大类间差法)将初始V通道图像二值化处理；
步骤S150：判断二值化图像中像素点的亮度强度值是否大于预设阈值；若二值化图像中像素点的亮度强度值大于预设阈值，则执行步骤S160；若二值化图像中像素点的亮度强度值不大于预设阈值，则执行步骤S170；
其中，所述二值化图像中像素点的亮度强度值或为0，或为255，用于识别 亮区像素点/暗区像素点的预设阈值的取值范围不受限制，0-255(包括0，但不包括255)的任意数值均可。二值化图像中定义二维坐标，每一像素点具有唯一的坐标值。按照预定的顺序遍历带有坐标值的二值化图像，直至将二值化图像中的所有像素点全部判断完成。
步骤S160：识别所述像素点为亮区像素点，并根据所述像素点的坐标值获得初始V通道图像的初始V通道特征值；
步骤S170：识别所述像素点为暗区像素点，并根据所述像素点的坐标值获得初始V通道图像的初始V通道特征值。
在本实用新型的另一个实施例中，预设若干亮度强度差值分段区间，通过亮度强度差值与亮度强度差值分段区间的识别获取不同的亮度补偿值，从而实现自适应的补偿效果。所述不同亮度补偿值可通过设置若干预设亮度补偿值(定值)获取，也可根据亮度强度差值按照预设算法获取。请参阅图3所示，在本实用新型的另一个实施例中，预设若干亮度强度差值分段区间，且所述步骤S30进一步包括以下步骤：
步骤S310：计算亮区平均像素亮度强度ValueB与暗区平均像素亮度强度ValueS的亮度强度差值DiffValue＝ValueB-ValueS；
步骤S320：识别亮度强度差值DiffValue所对应的亮度强度差值分段区间，并根据亮度强度差值分段区间选取相应的调整参数K；
步骤S330：根据所述亮度强度差值DiffValue、调整参数K及预设算法获取亮度补偿值OffsetValue，且所述预设算法为：OffsetValue＝K*DiffValue。
所述预设算法设置一个调整参数K，也可根据需要设置多个调整参数K1、K2，预设算法为：OffsetValue＝K1*DiffValue+K2，所述调整参数K1与DiffValue 关联，调整参数K2为定值参数，通过多个调整参数与DiffValue的组合调整，以便于取得更佳的补偿效果。
如表1进行，可根据需要设置亮度强度差值分段区间，通过亮度强度差值分段区间进行分段补偿，根据DiffValue的大小自动选择相应的调整参数K，具有自适应的补偿效果，其中，调整参数K的取值不限于以下数据。
DiffValue K OffsetValue
DiffValue≤valueA 0.5 OffsetValue＝0.5*DiffValue
ValueA<DiffValue≤valueB 1 OffsetValue＝1*DiffValue
ValueB<DiffValue≤valueC 1.2 OffsetValue＝1.2*DiffValue
ValueC<DiffValue 0.5 OffsetValue＝0.5*DiffValue
表1
亮度强度差值分段区间的数量及区间的端点值可根据需要而设定，例如，表1通过valueA、valueB、valueC、valueD为端点设置四个亮度强度差值分段区间，当DiffValue识别为其一亮度强度差值分段区间后，便可获知该亮度强度差值分段区间所对应的K，例如，当ValueB<DiffValue≤valueC时，选取K＝1.2，再由预设算法获取亮度补偿值。
在本实用新型的另一个实施例中，所述步骤S40根据亮度补偿值对亮区像素点和/或暗区像素点的初始V通道特征值进行补偿处理，例如，可只对亮区像素点的初始V通道特征值进行补偿处理，或者，只对暗区像素点的初始V通道特征值进行补偿处理，或者，同时对亮区像素点与暗区像素点的初始V通道特征值进行补偿处理的初始V通道特征值分别进行补偿处理。
亮区像素点：根据亮度补偿值对亮区像素点的初始V通道特征值进行减弱 补偿处理：补偿后V通道特征值＝初始V通道特征值-亮度补偿值。
暗区像素点：根据亮度补偿值对暗区像素点的初始V通道特征值进行增强补偿处理：补偿后V通道特征值＝初始V通道特征值+亮度补偿值。
请参阅图4所示，在本实用新型的另一个实施例中，所述步骤S50进一步包括：
步骤S510：根据H通道特征值、S通道特征值、补偿后V通道特征值进行合成处理，以获得去阴影边界的HSV图像(如图10所示)；
步骤S520：将去阴影HSV图像进行转换处理，以获得去阴影边界的RGB图像。
请参阅图5所示，在本实用新型的另一个实施例中，所述步骤S10与步骤S20之间，还包括：
步骤S15：判断所有像素点是否全为亮区像素点或暗区像素点；若是，则结束本次流程；若否，则执行步骤S20。
通过步骤S15筛选需要补偿和不需补偿的情形，从而优化补偿流程。
请参阅图6所示，在本实用新型的另一个实施例中，本发明提供一种图像处理方法；其包括以下步骤：
步骤S10：将原始图像进行预处理以识别每个像素点为亮区像素点或暗区像素点，并提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值；
步骤S16：根据S通道特征值筛选出预处理的像素点，其中，预处理的像素点中所有亮区像素点为第一像素点组合，预处理的像素点中所有暗区像素点为第二像素点组合；
步骤S20A：根据第一像素点组合的亮区像素点的初始V通道特征值计算亮区平均像素亮度强度ValueB，根据第二像素点组合的暗区像素点的初始V通道特征值计算暗区平均像素亮度强度ValueS；
步骤S30：根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue；
步骤S40：根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值；
步骤S50A：根据初始H通道特征值、初始S通道特征值、初始V通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像。
所述筛选条件可根据需要设定，像素点的S通道特征值大于预设S值satValue(satValue＝80)，则需对其进行补偿处理，像素点的S通道特征值不大于预设S值satValue(satValue＝80)，则不需对其进行补偿处理，通过步骤S16筛选处理以使部分图像中的V通道特征值不受补偿处理影响，即，补偿处理前后，部分图像仍保留初始V通道特征值。通过有选择地进行补偿处理，以避免部分图像的信息丢失。
假设原始图像的部分区域为车体，根据车体的S通道特征值的特征设置筛选条件，筛选处理后，会将车体以外的像素点作为预处理的像素点时，而将车体部分过滤掉，即补偿处理过程，不会对图像中的车体信息进行补偿，车体的像素点特征保持不变，因此，图像中的车体在去阴影前后保持不变。
请参阅图7所示，本发明还提供一种图像处理系统1，包括：
预处理模块10，其用于将原始图像进行预处理以识别每个像素点为亮区像 素点或暗区像素点，并用于提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值；
亮度补偿模块20，其用于根据亮区像素点的V通道特征值计算亮区平均像素亮度强度ValueB，根据暗区像素点的V通道特征值计算暗区平均像素亮度强度ValueS；并根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue；根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值；
图像生成模块30，其用于根据初始H通道特征值、初始S通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像。
本发明又提供一种自动行走设备，包括存储器和处理器，所述存储器存储有计算机程序，且所述处理器执行所述计算机程序时实现所述图像处理方法的步骤。本发明的自动行走设备可以是自动割草机，或者自动吸尘器等，也可以为其它设备，如喷洒设备、除雪设备、监视设备等等适合无人值守的设备。
本发明又提供一种可读存储介质，其上存储有计算机程序，且所述计算机程序被处理器执行时实现所述图像处理方法的步骤。
综上所述，本发明通过亮度补偿值调整V通道特征值以消除图像的阴影边界，避免因图像的阴影边界导致边界系统误判。本发明可将二值化图像中的像素点的亮度强度值与预设阈值进行比较处理，可识别亮区像素点/暗区像素点，且所述预设阈值的取值范围不受限制。本发明中设置的预设算法的计算量小，图片处理速度快。本发明通过设置亮度强度差值分段区间，通过亮度强度差值与亮度强度差值分段区间的识别获取不同的亮度补偿值，从而实现自适应的补偿效果。
此外，应当理解，虽然本说明书按照实施方式加以描述，但并非每个实施方式仅包含一个独立的技术方案，说明书的这种叙述方式仅仅是为清楚起见，本领域技术人员应当将说明书作为一个整体，各实施方式中的技术方案也可以经适当组合，形成本领域技术人员可以理解的其他实施方式。
上文所列出的一系列的详细说明仅仅是针对本发明的可行性实施方式的具体说明，并非用以限制本发明的保护范围，凡未脱离本发明技艺精神所作的等效实施方式或变更均应包含在本发明的保护范围之内。

Claims (10)

  1. 一种图像处理方法；其特征在于，包括以下步骤：
    将原始图像进行预处理以识别每个像素点为亮区像素点或暗区像素点，并提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值；
    根据亮区像素点的初始V通道特征值计算亮区平均像素亮度强度ValueB，根据暗区像素点的初始V通道特征值计算暗区平均像素亮度强度ValueS；
    根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue；
    根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值；
    根据初始H通道特征值、初始S通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像。
  2. 根据权利要求1所述的图像处理方法，其特征在于，所述将原始图像进行预处理的过程包括以下步骤：
    获取RGB图像；
    将RGB图像转换为HSV图像，提取HSV图像中每个像素点的特征值，所述特征值包括初始H通道特征值、初始S通道特征值与初始V通道特征值；
    根据HSV图像中每个像素点的初始V通道特征值以获得初始V通道图像；
    将初始V通道图像进行二值化处理，获得二值化图像；
    判断二值化图像中像素点的亮度强度值是否大于预设阈值，以识别每个像素点为亮区像素点或暗区像素点，二值化图像中像素点的亮度强度值大于预设阈值，则为亮区像素点，而二值化图像中像素点的亮度强度值不大于预设阈值，则为暗区像素点。
  3. 根据权利要求1所述的图像处理方法，其特征在于，在识别亮区像素点或暗区像素点后，还包括：
    判断所有像素点是否全为亮区像素点或暗区像素点；若是，则结束本次流程；若否，则继续执行后面的补偿流程。
  4. 根据权利要求1所述的图像处理方法，其特征在于，在识别亮区像素点或暗区像素点后，还包括：
    根据S通道特征值筛选出预处理的像素点，其中，预处理的像素点中所有亮区像素点为第一像素点组合，预处理的像素点中所有暗区像素点为第二像素点组合；
    则后续流程中，用于计算亮区平均像素亮度强度的亮区像素点为第一像素点组合的亮区像素点；用于计算暗区平均像素亮度强度的暗区像素点为第二像素点组合的暗区像素点。
  5. 根据权利要求1所述的图像处理方法，其特征在于，预设若干亮度强度差值分段区间，且所述获取亮度补偿值的过程包括：
    计算亮区平均像素亮度强度ValueB与暗区平均像素亮度强度ValueS的亮度强度差值DiffValue＝ValueB-ValueS；
    识别亮度强度差值DiffValue所对应的亮度强度差值分段区间，并根据亮度强度差值分段区间选取相应的调整参数K；
    根据所述亮度强度差值DiffValue、调整参数K及预设算法获取亮度补偿值OffsetValue，且所述预设算法为：OffsetValue＝K*DiffValue。
  6. 根据权利要求5所述的图像处理方法，其特征在于，根据亮度补偿值对亮区像素点和/或暗区像素点的初始V通道特征值进行补偿处理，根据亮度补偿 值对亮区像素点的初始V通道特征值进行减弱补偿处理：补偿后V通道特征值＝初始V通道特征值-亮度补偿值，根据亮度补偿值对暗区像素点的初始V通道特征值进行增强补偿处理：补偿后V通道特征值＝初始V通道特征值+亮度补偿值。
  7. 根据权利要求1所述的图像处理方法，其特征在于，所述合成处理为根据H通道特征值、S通道特征值、补偿后V通道特征值，以获得去阴影边界的HSV图像；所述转换处理为将去阴影HSV图像转换为去阴影边界的RGB图像。
  8. 一种图像处理系统，其特征在于，所述系统包括：
    预处理模块，其用于将原始图像进行预处理以识别每个像素点为亮区像素点或暗区像素点，并用于提取每个像素点的初始H通道特征值、初始S通道特征值和初始V通道特征值；
    亮度补偿模块，其用于根据亮区像素点的V通道特征值计算亮区平均像素亮度强度ValueB，根据暗区像素点的V通道特征值计算暗区平均像素亮度强度ValueS；并根据亮区平均像素亮度强度、暗区平均像素亮度强度获取亮度补偿值OffsetValue；根据亮度补偿值对初始V通道特征值进行补偿处理以获得补偿后V通道特征值；
    图像生成模块，其用于根据初始H通道特征值、初始S通道特征值和补偿后V通道特征值进行合成处理与转换处理以获得去阴影边界的图像。
  9. 一种自动行走设备，包括存储器和处理器，所述存储器存储有计算机程序，其特征在于，所述处理器执行所述计算机程序时实现权利要求1-7中任一项所述图像处理方法的步骤。
  10. 一种可读存储介质，其上存储有计算机程序，其特征在于，所述计算机 程序被处理器执行时实现权利要求1-7中任一项所述图像处理方法的步骤。
PCT/CN2020/115847 2020-03-27 2020-09-17 图像处理方法、系统、自动行走设备及可读存储介质 WO2021189782A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010227211.4A CN113449730A (zh) 2020-03-27 2020-03-27 图像处理方法、系统、自动行走设备及可读存储介质
CN202010227211.4 2020-03-27

Publications (1)

Publication Number Publication Date
WO2021189782A1 true WO2021189782A1 (zh) 2021-09-30

Family

ID=77807752

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2020/115847 WO2021189782A1 (zh) 2020-03-27 2020-09-17 图像处理方法、系统、自动行走设备及可读存储介质

Country Status (2)

Country Link
CN (1) CN113449730A (zh)
WO (1) WO2021189782A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116664613A (zh) * 2023-07-24 2023-08-29 合肥埃科光电科技股份有限公司 一种基于fpga的像素边缘位置检测方法、系统及存储介质
CN117059047A (zh) * 2023-10-12 2023-11-14 深圳市柯达科电子科技有限公司 一种用于lcd显示图像色彩智能调整方法

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114677336B (zh) * 2022-03-10 2024-04-09 四川省建筑科学研究院有限公司 一种基于红外图像的幕墙面板损伤识别方法
CN117237239B (zh) * 2023-11-14 2024-02-02 南京凯视迈科技有限公司 一种拼接图像的暗区消除方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101118721A (zh) * 2006-07-31 2008-02-06 三星电子株式会社 补偿阴影区域的方法、介质和系统
CN101236606A (zh) * 2008-03-07 2008-08-06 北京中星微电子有限公司 视频监控中的阴影消除方法及系统
CN104008528A (zh) * 2014-05-21 2014-08-27 河海大学常州校区 基于阈值分割的非均匀光场水下目标探测图像增强方法
CN104427318A (zh) * 2013-08-26 2015-03-18 Cjcgv株式会社 校正图像交叠区域的方法、记录介质和执行装置
CN107742511A (zh) * 2017-11-08 2018-02-27 颜色空间(北京)科技有限公司 显示屏拼接后模块间视觉缝隙消除的方法以及显示方法
CN108711140A (zh) * 2018-05-16 2018-10-26 广东欧谱曼迪科技有限公司 一种基于类间方差描述的图像亮度均匀性实时恢复方法
CN108830800A (zh) * 2018-05-09 2018-11-16 南京邮电大学 一种暗光场景下图像的亮度提升增强方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101118721A (zh) * 2006-07-31 2008-02-06 三星电子株式会社 补偿阴影区域的方法、介质和系统
CN101236606A (zh) * 2008-03-07 2008-08-06 北京中星微电子有限公司 视频监控中的阴影消除方法及系统
CN104427318A (zh) * 2013-08-26 2015-03-18 Cjcgv株式会社 校正图像交叠区域的方法、记录介质和执行装置
CN104008528A (zh) * 2014-05-21 2014-08-27 河海大学常州校区 基于阈值分割的非均匀光场水下目标探测图像增强方法
CN107742511A (zh) * 2017-11-08 2018-02-27 颜色空间(北京)科技有限公司 显示屏拼接后模块间视觉缝隙消除的方法以及显示方法
CN108830800A (zh) * 2018-05-09 2018-11-16 南京邮电大学 一种暗光场景下图像的亮度提升增强方法
CN108711140A (zh) * 2018-05-16 2018-10-26 广东欧谱曼迪科技有限公司 一种基于类间方差描述的图像亮度均匀性实时恢复方法

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116664613A (zh) * 2023-07-24 2023-08-29 合肥埃科光电科技股份有限公司 一种基于fpga的像素边缘位置检测方法、系统及存储介质
CN116664613B (zh) * 2023-07-24 2023-10-31 合肥埃科光电科技股份有限公司 一种基于fpga的像素边缘位置检测方法、系统及存储介质
CN117059047A (zh) * 2023-10-12 2023-11-14 深圳市柯达科电子科技有限公司 一种用于lcd显示图像色彩智能调整方法
CN117059047B (zh) * 2023-10-12 2023-12-22 深圳市柯达科电子科技有限公司 一种用于lcd显示图像色彩智能调整方法

Also Published As

Publication number Publication date
CN113449730A (zh) 2021-09-28

Similar Documents

Publication Publication Date Title
WO2021189782A1 (zh) 图像处理方法、系统、自动行走设备及可读存储介质
US9842382B2 (en) Method and device for removing haze in single image
CN107862667B (zh) 一种基于高分辨率遥感影像的城市阴影检测与去除方法
CN109087254B (zh) 无人机航拍图像雾霾天空和白色区域自适应处理方法
CN113077486B (zh) 一种山区植被覆盖率监测方法及系统
CN105631823A (zh) 基于阈值分割优化的暗原色天空区域去雾方法
CN108875747A (zh) 一种基于机器视觉的小麦不完善粒识别方法
CN106096491B (zh) 一种眼底彩色照相图像中微动脉瘤自动化识别方法
CN111681253A (zh) 基于颜色与形态特征的叶片图像分割方法及系统
CN103336966A (zh) 一种应用于农业智能机械的杂草图像辨识方法
CN110909686B (zh) 一种用于辅助驾驶的低照度图像增强系统
CN107067375A (zh) 一种基于暗通道先验和边缘信息的图像去雾方法
CN111145105B (zh) 一种图像快速去雾方法、装置、终端及存储介质
CN112053298A (zh) 一种图像去雾方法
CN110188640B (zh) 人脸识别方法、装置、服务器和计算机可读介质
CN109712116B (zh) 输电线路及其附件的故障识别方法
CN111046782A (zh) 一种用于苹果采摘机器人的果实快速识别方法
CN112561919A (zh) 一种图像分割方法、设备和计算机可读存储介质
CN116843581B (zh) 一种多场景图的图像增强方法、系统、设备和存储介质
CN111667509B (zh) 目标与背景颜色相似下的运动目标自动跟踪方法及系统
US11354794B2 (en) Deposit detection device and deposit detection method
CN113095283A (zh) 一种基于动态roi和改进萤火虫算法的车道线提取方法
CN117274085A (zh) 低照度图像增强方法及装置
CN116363097A (zh) 一种光伏板的缺陷检测方法及系统
CN115564682A (zh) 一种光照不均图像增强方法及系统

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 20926458

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 20926458

Country of ref document: EP

Kind code of ref document: A1