WO2020107662A1 - 多曝光图像融合方法 - Google Patents

多曝光图像融合方法 Download PDF

Info

Publication number
WO2020107662A1
WO2020107662A1 PCT/CN2019/070915 CN2019070915W WO2020107662A1 WO 2020107662 A1 WO2020107662 A1 WO 2020107662A1 CN 2019070915 W CN2019070915 W CN 2019070915W WO 2020107662 A1 WO2020107662 A1 WO 2020107662A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
pixel
exposure
exposure image
row
Prior art date
Application number
PCT/CN2019/070915
Other languages
English (en)
French (fr)
Inventor
史超超
Original Assignee
深圳市华星光电半导体显示技术有限公司
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 深圳市华星光电半导体显示技术有限公司 filed Critical 深圳市华星光电半导体显示技术有限公司
Publication of WO2020107662A1 publication Critical patent/WO2020107662A1/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
    • G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10024—Color image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20212—Image combination
    • G06T2207/20221—Image fusion; Image merging

Definitions

  • the invention relates to the field of image processing, in particular to a multi-exposure image fusion method.
  • Thin film transistor Thin Film Transistor, TFT is the main driving element in the current liquid crystal display device (Liquid Crystal Display) and active matrix driven organic electroluminescence display device (Active Matrix Organic Light-Emitting Diode, AMOLED), It is directly related to the display performance of the flat panel display device.
  • liquid crystal displays which include a liquid crystal display panel and a backlight module.
  • the working principle of the liquid crystal display panel is to infuse liquid crystal molecules between the thin film transistor array substrate (Thin Film Transistor Array Substrate, TFT Array Substrate) and the color filter (CF) substrate, and apply them separately on the two substrates
  • the pixel voltage and the common voltage control the rotation direction of the liquid crystal molecules by the electric field formed between the pixel voltage and the common voltage, so as to transmit the light of the backlight module to generate a picture.
  • multi-exposure image fusion is needed to integrate multiple images with different exposure levels to obtain images with higher information content.
  • the existing multi-exposure fusion algorithm generates multiple exposure images from the original image by constructing an appropriate exposure function.
  • the multiple exposure images calculate their respective weights, and in turn, the respective weights merge the multiple exposure images into the target image.
  • each exposure image is obtained by taking the average value of the original image as the central value, and multiple exposure images are compared with the central value to obtain the weight, and each different exposure image tends to focus on different points, such as darker
  • the exposure image of is often focused on the brightest area (such as the sky).
  • the brightest exposure image needs to be enhanced in the details of the darker area.
  • the uniform weighting of each exposure image can not achieve good results.
  • the fused image is often whitish or blurry.
  • An object of the present invention is to provide a multi-exposure image fusion method, which can enhance the details of the target image after the image fusion and prevent the target image after the image fusion from being whitish or blurred.
  • the present invention provides a multi-exposure image fusion method, including the following steps:
  • Step S1 Extract the brightness component of the original image, and use the S-type function to generate K exposure images according to the brightness component, and let K be a positive integer;
  • Step S2 Calculate the weight of each exposure image according to the average brightness of the area to be enhanced of each exposure image
  • Step S3 Select the corresponding image fusion formula according to the distribution type of the cumulative histogram of the original image, and obtain the brightness value of the target image according to the weight of each exposure image and the image fusion formula.
  • the S-shaped function is: , Where Lwk(i, j) is the brightness value of the pixel in the i-th row and j-th column of the K-th exposure image, 10-pk is the scaling factor of the K-th exposure image, and Lad, k is the The average brightness, Lmax,k is the maximum brightness of the Kth exposure image.
  • the weight calculation formula is: Among them, Wk(i,j) is the weight of the pixels in the i-th row and j-th column of the K-th exposure image, and Lmed,k is the average value of the brightness of the area to be enhanced in the K-th exposure image.
  • step S3 the method of determining the distribution type of the cumulative histogram of the original image is: generating K-1 probability distribution blocks corresponding to all pixels of the original image according to the cumulative histogram of the original image, and each probability distribution The difference between the maximum gray level and the minimum gray level of the block is equal;
  • the distribution type of the cumulative histogram of the original image is determined to be a distribution type at both ends.
  • the preset cumulative probability sum is 0.65.
  • the image fusion formula corresponding to the distributed at both ends is: Where L(i,j) is the brightness value of the pixel in row i and column j of the target image, and L wn (i,j) is the brightness value of the pixel in row i and column j of the first exposure image to the Kth frame The brightness value of the pixel in the i-th row and j-th column of the exposure image, W n (i,j) is the weight of the pixel in the i-th row and j-th column of the first exposure image to the i-th row and j-th column of the Kth exposure image The weight of the pixel.
  • the image fusion formula corresponding to the intermediate distribution type is: Where L(i,j) is the brightness value of the pixel in row i and column j of the target image, and L wn (i,j) is the brightness value of the pixel in row i and column j of the first exposure image to the Kth frame The brightness value of the pixel in the i-th row and j-th column of the exposure image, W k-n+1 (i,j) is the weight of the pixel in the i-th row and j-th column of the K -th exposure image to the i-th of the first exposure image The weight of the pixel in row j.
  • the gray scale of each pixel of each exposure image is also reduced by 255 times.
  • the invention also provides a multi-exposure image fusion method, including the following steps:
  • Step S1 Extract the brightness component of the original image, and use the S-type function to generate K exposure images according to the brightness component, and let K be a positive integer;
  • Step S2 Calculate the weight of each exposure image according to the average brightness of the area to be enhanced of each exposure image
  • Step S3 Select the corresponding image fusion formula according to the distribution type of the cumulative histogram of the original image, and obtain the brightness value of the target image according to the weight of each exposure image and the image fusion formula;
  • the gray scale of each pixel of each exposure image is also reduced by 255 times.
  • the multi-exposure image fusion method of the present invention by extracting the luminance component of the original image, using the S-type function to generate K exposure images according to the luminance component, and calculating each image based on the average brightness of the area to be enhanced for each exposure image
  • the weight of the exposed image select the corresponding image fusion formula according to the distribution type of the cumulative histogram of the original image, and substitute the weight of each exposure image into the image fusion formula to obtain the brightness of the target image to enhance the details of the target image after image fusion, Prevent the target image after image fusion from whitish or blurry.
  • FIG. 1 is a flowchart of a multi-exposure image fusion method of the present invention.
  • the present invention provides a multi-exposure image fusion method, including the following steps:
  • Step S1 Extract the brightness component of the original image, and use the S-type function to generate K exposure images according to the brightness component, and let K be a positive integer;
  • Step S2 Calculate the weight of each exposure image according to the average brightness of the area to be enhanced of each exposure image
  • Step S3 Select the corresponding image fusion formula according to the distribution type of the cumulative histogram of the original image, and obtain the brightness value of the target image according to the weight of each exposure image and the image fusion formula.
  • the exposure values of the K exposure images are sequentially increased, that is, the exposure value of the first exposure image is the smallest, and the exposure value of the K exposure image is the largest.
  • the S-shaped function is: , Where L wk (i, j) is the brightness value of the pixel in the i-th row and j-th column of the K-th exposure image, 10 -p k is the scaling factor of the K-th exposure image, and L ad,k is the K-th exposure image The average brightness of the exposed image, L max,k is the maximum brightness of the Kth exposure image.
  • step S1 the gray scale of each pixel of each exposure image can also be reduced by 255 times, that is, from 0 to 255 gray scales are compressed to 0-1 gray scales. From the histogram, it will be gray The abscissa of the order is compressed from 0-255 to 0-1 for easy statistics.
  • step S2 the weight calculation formula is: Among them, W k (i, j) is the weight of the pixels in the i-th row and j-th column of the K-th exposure image, and L med, k is the average value of the region brightness to be enhanced in the K-th exposure image.
  • the method of determining the distribution type of the cumulative histogram of the original image is: generating K-1 probability distribution blocks corresponding to all pixels of the original image according to the cumulative histogram of the original image, and each The difference between the maximum gray level and the minimum gray level of a probability distribution block is the same; when the K-1 equal probability distribution blocks except the first probability distribution block and the K-1 probability distribution block The sum of the cumulative probabilities of multiple adjacent probability distribution blocks is greater than a preset sum of cumulative probabilities, and the distribution type of the cumulative histogram of the original image is determined to be an intermediate distribution type, when K-1 equal probability distributions Except for the first probability distribution block and the K-1th probability distribution block, the sum of the cumulative probabilities of multiple adjacent probability distribution blocks in the block is less than or equal to a predetermined cumulative probability And, it is judged that the distribution type of the cumulative histogram of the original image is a distribution type at both ends.
  • the preset cumulative probability sum is 0.65.
  • the image fusion formula corresponding to the two-end distribution type is: Where L(i,j) is the brightness value of the pixel in row i and column j of the target image, and L wn (i,j) is the brightness value of the pixel in row i and column j of the first exposure image to the Kth frame The brightness value of the pixel in the i-th row and j-th column of the exposure image, W n (i,j) is the weight of the pixel in the i-th row and j-th column of the first exposure image to the i-th row and j-th column of the Kth exposure image The weight of the pixel.
  • the image fusion formula corresponding to the distribution at both ends is the product of the brightness value of the pixel in the i-th row and j-th column of the first exposure image and the weight of the pixel in the i-th row and j-th column of the first exposure image, plus the The product of the brightness value of the pixel in row i and column j of the two exposure images and the weight of the pixel in row i and column j of the second exposure image until the pixel in row i and column j of the K exposure image is added Multiplied by the brightness value of the pixel in row i and column j of the Kth exposure image to obtain the brightness value of the pixel in row i and column j of the target image.
  • the The exposure value of the K exposure images is the highest, which can enhance the details of the bright areas of the exposure images with low exposure values and the dark areas of the exposure images with high exposure values, thereby enhancing the details of the target image after image fusion.
  • the image fusion formula corresponding to the intermediate distribution type is: Where L(i,j) is the brightness value of the pixel in row i and column j of the target image, and L wn (i,j) is the brightness value of the pixel in row i and column j of the first exposure image to the Kth frame The brightness value of the pixel in the i-th row and j-th column of the exposure image, W k-n+1 (i,j) is the weight of the pixel in the i-th row and j-th column of the K -th exposure image to the i-th of the first exposure image The weight of the pixel in row j.
  • the image fusion formula corresponding to the intermediate distribution type is the product of the brightness value of the pixel in the ith row and jth column of the first exposure image and the weight of the pixel in the ith row and jth column of the Kth exposure image, plus the second The product of the brightness values of the pixels in the i-th row and j-th column of the exposure image and the weights of the pixels in the i-th row and j-th column of the K-1 exposure image, until the i-th row and j-th column of the K-th exposure image.
  • the product of the brightness value of the pixel and the weight of the pixel in the i-th row and j-th column of the first exposure image to obtain the brightness value of the pixel in the i-th row and j-th column of the target image, since the exposure value of the first exposure image is the lowest,
  • the exposure value of the Kth exposure image is the highest, and the enhancement weight can be changed to make the low exposure value of the low exposure value and the high exposure value
  • Step S1 generates 5 exposure images, the exposure value of the first exposure image is the smallest, and the exposure value of the fifth exposure image is the largest, and Step S2 corresponds to generating 4 equal probability distribution blocks, The difference between the maximum gray level and the minimum gray level of each probability distribution block is equal, that is, the 4 probability distribution blocks are divided into 255 gray levels, the gray range of the first probability distribution block is 0-64, the second The gray-scale range of the probability distribution block is 64-128, the gray-scale range of the third probability distribution block is 128-192, and the gray-scale range of the fourth probability distribution block is 192-255.
  • Step S3 determines the second Whether the sum of the cumulative probability of the probability distribution blocks and the cumulative probability of the third probability distribution block is greater than 0.65 (that is, whether the ratio of the number of pixels located in the 64-192 gray scale to the total number of pixels is greater than 0.65),
  • the distribution type of the cumulative histogram of the original image is judged to be a middle distribution type, and when the sum of the cumulative probabilities is less than or equal to 0.65, the distribution type of the cumulative histogram of the original image is judged to be a distribution type at both ends
  • the multi-exposure image fusion method of the present invention by extracting the brightness component of the original image, an S-shaped function is used to generate K exposure images, and each exposure is calculated according to the average brightness of the area to be enhanced for each exposure image
  • the weight of the image select the corresponding image fusion formula according to the distribution type of the cumulative histogram of the original image, and substitute the weight of each exposure image into the image fusion formula to obtain the brightness of the target image to enhance the details of the target image after image fusion and prevent
  • the target image after image fusion is whitish or blurred.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)
  • Facsimile Image Signal Circuits (AREA)
  • Studio Devices (AREA)

Abstract

本发明提供一种多曝光图像融合方法。该多曝光图像融合方法通过提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；根据原始图像的累积直方图的分布类型选择对应的图像融合公式，将每张曝光图像的权重代入图像融合公式获得目标图像的亮度，以增强图像融合后的目标图像的细节，防止图像融合后的目标图像发白或模糊。

Description

多曝光图像融合方法 技术领域
本发明涉及图像处理领域，尤其涉及一种多曝光图像融合方法。
背景技术
薄膜晶体管(Thin Film Transistor，TFT)是目前液晶显示装置(Liquid Crystal Display，LCD)和有源矩阵驱动式有机电致发光显示装置(Active Matrix Organic Light-Emitting Diode，AMOLED)中的主要驱动元件，直接关系平板显示装置的显示性能。
现有市场上的液晶显示器大部分为背光型液晶显示器，其包括液晶显示面板及背光模组(backlight module)。液晶显示面板的工作原理是在薄膜晶体管阵列基板(Thin Film Transistor Array Substrate，TFT Array Substrate)与彩色滤光片(Color Filter，CF)基板之间灌入液晶分子，并在两片基板上分别施加像素电压和公共电压，通过像素电压和公共电压之间形成的电场控制液晶分子的旋转方向，以将背光模组的光线透射出来产生画面。
由于在同一场景不同光线下得到的图像，无论它的曝光时间长短，都会出现曝光过度或曝光不足的现象，很容易在图像中产生阴影及光照不均等现象，这样造成图像信息含量低，重要信息丢失等问题。因此需要多曝光图像融合将多张不同曝光程度的图像加以综合，以得到信息含量较高的图像。现有的多曝光融合算法通过构建一个合适的曝光函数来将原始图像生成多张曝光图像，多张曝光图像分别计算各自的权重，依次各自的权重将多张曝光图像融合成目标图像。但是每张曝光图像的权重计算求取得是原始图像的均值作为中心值，多张曝光图像分别与该中心值进行比较求得权重，而每张不同的曝光图像往往关注的重点不同，例如较暗的曝光图像往往关注的是最亮的那个区域(如天空)，最亮的曝光图像需要增强的地方是较暗的区域细节，每张曝光图像的权重统一的取均值并不能取得较好的效果，融合后的图像往往发白或模糊。
发明内容
本发明的目的在于提供一种多曝光图像融合方法，能够增强图像融合后的目标图像的细节，防止图像融合后的目标图像发白或模糊。
为实现上述目的，本发明提供了一种多曝光图像融合方法，包括如下 步骤：
步骤S1、提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，设K为正整数；
步骤S2、根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；
步骤S3、根据原始图像的累积直方图的分布类型选择对应的图像融合公式，根据每张曝光图像的权重以及图像融合公式获得目标图像的亮度值。
所述步骤S1中，提取原始图像的亮度分量的公式为：Ld(i,j)＝0.2125*R(i,j)+0.7154*G(i,j)+0.0721*B(i,j)；其中，设i，j均为正整数，Ld(i,j)为原始图像的第i行第j列像素的亮度值，R(i,j)为原始图像的第i行第j列像素的红色子像素的亮度值，G(i,j)为原始图像的第i行第j列像素的绿色子像素的亮度值，B(i,j)为原始图像的第i行第j列像素的蓝色子像素的亮度值。
所述S型函数为：
Figure PCTCN2019070915-appb-000001
，其中，Lwk(i，j)为第K张曝光图像的第i行第j列像素的亮度值，10-pk为第K张曝光图像的缩放因子，Lad,k为第K张曝光图像的平均亮度，Lmax,k为第K张曝光图像的最大亮度。
第K张曝光图像的平均亮度的计算公式为：L ad,k＝1+exp(μ*V k)，其中，μ为常数，Vk为第K张曝光图像的曝光值。
所述步骤S2中，权重计算公式为：
Figure PCTCN2019070915-appb-000002
其中，Wk(i,j)为第K张曝光图像的第i行第j列像素的权重，Lmed,k为第K张曝光图像需增强区域亮度均值。
所述步骤S3中，判断原始图像的累积直方图的分布类型的方法为：根据原始图像的累积直方图生成对应原始图像的所有像素的K-1个的概率分 布区块，且每一个概率分布区块的最大灰阶与最小灰阶之差相等；
当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中的多个相邻的概率分布区块的累积概率之和大于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为中间分布型；
当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中多个相邻的概率分布区块的累积概率之和小于或等于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为两端分布型。
预设的累积概率之和为0.65。
当原始图像的累积直方图的类型为两端分布型时，对应两端分布型的图像融合公式为：
Figure PCTCN2019070915-appb-000003
其中L(i,j)为目标图像的第i行第j列像素的亮度值，L wn(i,j)为第1张曝光图像的第i行第j列像素的亮度值至第K张曝光图像的第i行第j列像素的亮度值，W n(i,j)为第1张曝光图像的第i行第j列像素的权重至第K张曝光图像的第i行第j列像素的权重。
当原始图像的累积直方图的分布类型为中间分布型时，对应中间分布型的图像融合公式为：
Figure PCTCN2019070915-appb-000004
其中L(i,j)为目标图像的第i行第j列像素的亮度值，L wn(i,j)为第1张曝光图像的第i行第j列像素的亮度值至第K张曝光图像的第i行第j列像素的亮度值，W k-n+1(i,j)为第K张曝光图像的第i行第j列像素的权重至第1张曝光图像的第i行第j列像素的权重。
所述步骤S1中还将每张曝光图像的每一像素的灰阶缩小255倍。
本发明还提供了一种多曝光图像融合方法，包括如下步骤：
步骤S1、提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，设K为正整数；
步骤S2、根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；
步骤S3、根据原始图像的累积直方图的分布类型选择对应的图像融合公式，根据每张曝光图像的权重以及图像融合公式获得目标图像的亮度值；
所述步骤S1中，提取原始图像的亮度分量的公式为：Ld(i,j)＝0.2125*R(i,j)+0.7154*G(i,j)+0.0721*B(i,j)；其中，设i，j均为正整数，Ld(i,j)为原始图像的第i行第j列像素的亮度值，R(i,j)为原始图像的第i行第j列像素的红色子像素的亮度值，G(i,j)为原始图像的第i行第j列像素的绿色子像素的亮度值，B(i,j)为原始图像的第i行第j列像素的蓝色子像素的亮度值；
所述步骤S1中还将每张曝光图像的每一像素的灰阶缩小255倍。
本发明的有益效果：本发明的多曝光图像融合方法，通过提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；根据原始图像的累积直方图的分布类型选择对应的图像融合公式，将每张曝光图像的权重代入图像融合公式获得目标图像的亮度，以增强图像融合后的目标图像的细节，防止图像融合后的目标图像发白或模糊。
附图说明
为了能更进一步了解本发明的特征以及技术内容，请参阅以下有关本发明的详细说明与附图，然而附图仅提供参考与说明用，并非用来对本发明加以限制。
附图中，
图1为本发明的多曝光图像融合方法的流程图。
具体实施方式
为更进一步阐述本发明所采取的技术手段及其效果，以下结合本发明的优选实施例及其附图进行详细描述。
请参阅图1，本发明提供一种多曝光图像融合方法，包括如下步骤：
步骤S1、提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，设K为正整数；
步骤S2、根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；
步骤S3、根据原始图像的累积直方图的分布类型选择对应的图像融合公式，根据每张曝光图像的权重以及图像融合公式获得目标图像的亮度值。
具体的，所述K张曝光图像的曝光值依次递增，即第1张曝光图像的曝光值最小，第K张曝光图像的曝光值最大。
具体的，所述步骤S1中，提取原始图像的亮度分量的公式为：L d(i,j)＝ 0.2125*R(i.j)+0.7154*G(i,j)+0.0721*B(i,j)；其中，设i，j均为正整数，L d(i,j)为原始图像的第i行第j列像素的亮度值，R(i.j)为原始图像的第i行第j列像素的红色子像素的亮度值，G(i.j)为原始图像的第i行第j列像素的绿色子像素的亮度值，B(i.j)为原始图像的第i行第j列像素的蓝色子像素的亮度值。
具体的，所述S型函数为：
Figure PCTCN2019070915-appb-000005
，其中，L wk(i，j)为第K张曝光图像的第i行第j列像素的亮度值，10 -p k为第K张曝光图像的缩放因子，L ad,k为第K张曝光图像的平均亮度，L max,k为第K张曝光图像的最大亮度。
进一步的，第K张曝光图像的平均亮度的计算公式为：L ad,k＝1+exp(μ*V k)，其中，μ为常数，V k为第K张曝光图像的曝光值。
具体的，所述步骤S1中还可以将每张曝光图像的每一像素的灰阶缩小255倍，即将0到255灰阶压缩至0-1灰阶，从直方图来看，就是将为灰阶的横坐标从0-255压缩至0-1，以便于统计。
具体的，所述步骤S2中，权重计算公式为：
Figure PCTCN2019070915-appb-000006
其中，W k(i,j)为第K张曝光图像的第i行第j列像素的权重，L med,k为第K张曝光图像需增强区域亮度均值。
具体的，所述步骤S3中，判断原始图像的累积直方图的分布类型的方法为：根据原始图像的累积直方图生成对应原始图像的所有像素的K-1个的概率分布区块，且每一个概率分布区块的最大灰阶与最小灰阶之差相等；当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中的多个相邻的概率分布区块的累积概率之和大于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为中间分布型， 当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中的多个相邻的概率分布区块的累积概率之和小于或等于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为两端分布型。
具体的，预设的累积概率之和为0.65。
进一步的，当原始图像的累积直方图的类型为两端分布型时，对应两端分布型的图像融合公式为：
Figure PCTCN2019070915-appb-000007
其中L(i,j)为目标图像的第i行第j列像素的亮度值，L wn(i,j)为第1张曝光图像的第i行第j列像素的亮度值至第K张曝光图像的第i行第j列像素的亮度值，W n(i,j)为第1张曝光图像的第i行第j列像素的权重至第K张曝光图像的第i行第j列像素的权重。即对应两端分布型的图像融合公式是将第1张曝光图像的第i行第j列像素的亮度值与第1张曝光图像的第i行第j列像素的权重的乘积，加上第2张曝光图像的第i行第j列像素的亮度值与第2张曝光图像的第i行第j列像素的权重的乘积，直至加上第K张曝光图像的第i行第j列像素的亮度值与第K张曝光图像的第i行第j列像素的权重的乘积，从而得到目标图像的第i行第j列像素的亮度值，由于第1张曝光图像的曝光值最低，第K张曝光图像的的曝光值最高，可以增强低曝光值的曝光图像的亮区细节以及增强高曝光值的曝光图像的暗区细节，进而增强图像融合后的目标图像的细节。
当原始图像的累积直方图的分布类型为中间分布型时，对应中间分布型的图像融合公式为：
Figure PCTCN2019070915-appb-000008
其中L(i,j)为目标图像的第i行第j列像素的亮度值，L wn(i,j)为第1张曝光图像的第i行第j列像素的亮度值至第K张曝光图像的第i行第j列像素的亮度值，W k-n+1(i,j)为第K张曝光图像的第i行第j列像素的权重至第1张曝光图像的第i行第j列像素的权重。即对应中间分布型的图像融合公式是将第1张曝光图像的第i行第j列像素的亮度值与第K张曝光图像的第i行第j列像素的权重的乘积，加上第2张曝光图像的第i行第j列像素的亮度值与第K-1张曝光图像的第i行第j列像素的权重的乘积，直至加上第K张曝光图像的第i行第j列像素的亮度值与第1张曝光图像的第i行第j列像素的权重的乘积，从而得到目标图像的第i行第j列像素的亮度值，由于第1张曝光图 像的曝光值最低，第K张曝光图像的的曝光值最高，可以改变增强权重，从而使低曝光值的曝光图像的低曝，高曝光值的曝光图像的高曝，以增强低曝光值和高曝光值的曝光图像的细节，进而增强图像融合后的目标图像的细节。
以下以K＝5举例说明：步骤S1生成了5张曝光图像，第1张曝光图像的曝光值最小，第5张曝光图像的曝光值最大，步骤S2对应生成4个相等的概率分布区块，每一个概率分布区块的最大灰阶与最小灰阶之差相等，即4个概率分布区块均分255灰阶，第1个概率分布区块的灰阶范围为0-64，第2个概率分布区块的灰阶范围为64-128，第3个概率分布区块的灰阶范围为128-192，第4个概率分布区块的灰阶范围为192-255，步骤S3判断第2个概率分布区块的累积概率与第3个概率分布区块的累积概率之和是否大于0.65(也就是判断位于64-192灰阶的像素个数占总像素个数的比值是否大于0.65)，当累积概率之和大于0.65，判断原始图像的累积直方图的分布类型为中间分布型，当累积概率之和小于或等于0.65，判断原始图像的累积直方图的分布类型为两端分布型，当原始图像的累积直方图的类型为两端分布型时，对应两端分布型的图像融合公式为：L(i,j)＝L w1(i,j)* W1(i,j)+L w2(i,j)*W 2(i,j)+L w3(i,j)*W 3(i,j)+L w4(i,j)*W 4(i,j)+L w5(i,j)*W 5(i,j)，当原始图像的累积直方图的类型为中间分布型时，对应中间分布型的图像融合公式为：L(i,j)＝L w1(i,j)*W 5(i,j)+L w2(i,j)*W 4(i,j)+L w3(i,j)*W 3(i,j)+L w4(i,j)*W 2(i,j)+L w5(i,j)*W 1(i,j)，从而得到目标图像的第i行第j列像素的亮度值，以增强图像融合后的目标图像的细节，防止图像融合后的目标图像发白或模糊。
综上所述，本发明的多曝光图像融合方法，通过提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；根据原始图像的累积直方图的分布类型选择对应的图像融合公式，将每张曝光图像的权重代入图像融合公式获得目标图像的亮度，以增强图像融合后的目标图像的细节，防止图像融合后的目标图像发白或模糊。
以上所述，对于本领域的普通技术人员来说，可以根据本发明的技术方案和技术构思作出其他各种相应的改变和变形，而所有这些改变和变形都应属于本发明权利要求的保护范围。

Claims (15)

  1. 一种多曝光图像融合方法，包括如下步骤：
    步骤S1、提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，设K为正整数；
    步骤S2、根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；
    步骤S3、根据原始图像的累积直方图的分布类型选择对应的图像融合公式，根据每张曝光图像的权重以及图像融合公式获得目标图像的亮度值。
  2. 如权利要求1所述的多曝光图像融合方法，其中，所述步骤S1中，提取原始图像的亮度分量的公式为：Ld(i,j)＝0.2125*R(i,j)+0.7154*G(i,j)+0.0721*B(i,j)；其中，设i，j均为正整数，Ld(i,j)为原始图像的第i行第j列像素的亮度值，R(i,j)为原始图像的第i行第j列像素的红色子像素的亮度值，G(i,j)为原始图像的第i行第j列像素的绿色子像素的亮度值，B(i,j)为原始图像的第i行第j列像素的蓝色子像素的亮度值。
  3. 如权利要求2所述的多曝光图像融合方法，其中，所述S型函数为：
    Figure PCTCN2019070915-appb-100001
    ，其中，L wk(i，j)为第K张曝光图像的第i行第j列像素的亮度值，10 -p k为第K张曝光图像的缩放因子，L ad,k为第K张曝光图像的平均亮度，L max,k为第K张曝光图像的最大亮度。
  4. 如权利要求3所述的多曝光图像融合方法，其中，第K张曝光图像的平均亮度的计算公式为：L ad,k＝1+exp(μ*V k)，其中，μ为常数，V k为第K张曝光图像的曝光值。
  5. 如权利要求3所述的多曝光图像融合方法，其中，所述步骤S2中，权重计算公式为：
    Figure PCTCN2019070915-appb-100002
    其中，W k(i,j)为第K张曝光图像的第i行第j列像素的权重，L med,k为第K张曝光图像需增强区域亮度均值。
  6. 如权利要求5所述的多曝光图像融合方法，其中，所述步骤S3中，判断原始图像的累积直方图的分布类型的方法为：根据原始图像的累积直方图生成对应原始图像的所有像素的K-1个的概率分布区块，且每一个概率分布区块的最大灰阶与最小灰阶之差相等；
    当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中的多个相邻的概率分布区块的累积概率之和大于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为中间分布型；
    当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中多个相邻的概率分布区块的累积概率之和小于或等于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为两端分布型。
  7. 如权利要求6所述的多曝光图像融合方法，其中，预设的累积概率之和为0.65。
  8. 如权利要求6所述的多曝光图像融合方法，其中，当原始图像的累积直方图的类型为两端分布型时，对应两端分布型的图像融合公式为：
    Figure PCTCN2019070915-appb-100003
    其中L(i,j)为目标图像的第i行第j列像素的亮度值，L wn(i,j)为第1张曝光图像的第i行第j列像素的亮度值至第K张曝光图像的第i行第j列像素的亮度值，W n(i,j)为第1张曝光图像的第i行第j列像素的权重至第K张曝光图像的第i行第j列像素的权重。
  9. 如权利要求6所述的多曝光图像融合方法，其中，当原始图像的累积直方图的分布类型为中间分布型时，对应中间分布型的图像融合公式为：
    Figure PCTCN2019070915-appb-100004
    其中L(i,j)为目标图像的第i行第j列像素的亮度值，L wn(i,j)为第1张曝光图像的第i行第j列像素的亮度 值至第K张曝光图像的第i行第j列像素的亮度值，W k-n+1(i,j)为第K张曝光图像的第i行第j列像素的权重至第1张曝光图像的第i行第j列像素的权重。
  10. 如权利要求1所述的多曝光图像融合方法，其中，所述步骤S1中还将每张曝光图像的每一像素的灰阶缩小255倍。
  11. 一种多曝光图像融合方法，包括如下步骤：
    步骤S1、提取原始图像的亮度分量，根据亮度分量采用S型函数生成K张曝光图像，设K为正整数；
    步骤S2、根据每张曝光图像的需增强区域亮度均值计算每张曝光图像的权重；
    步骤S3、根据原始图像的累积直方图的分布类型选择对应的图像融合公式，根据每张曝光图像的权重以及图像融合公式获得目标图像的亮度值；
    其中，所述步骤S1中，提取原始图像的亮度分量的公式为：Ld(i,j)＝0.2125*R(i,j)+0.7154*G(i,j)+0.0721*B(i,j)；其中，设i，j均为正整数，Ld(i,j)为原始图像的第i行第j列像素的亮度值，R(i,j)为原始图像的第i行第j列像素的红色子像素的亮度值，G(i,j)为原始图像的第i行第j列像素的绿色子像素的亮度值，B(i,j)为原始图像的第i行第j列像素的蓝色子像素的亮度值；
    其中，所述步骤S1中还将每张曝光图像的每一像素的灰阶缩小255倍。
  12. 如权利要求11所述的多曝光图像融合方法，其中，所述S型函数为：
    Figure PCTCN2019070915-appb-100005
    ，其中，L wk(i，j)为第K张曝光图像的第i行第j列像素的亮度值，10 -p k为第K张曝光图像的缩放因子，L ad,k为第K张曝光图像的平均亮度，L max,k为第K张曝光图像的最大亮度。
  13. 如权利要求12所述的多曝光图像融合方法，其中，第K张曝光图像的平均亮度的计算公式为：L ad,k＝1+exp(μ*V k)，其中，μ为常数，V k为第K张曝光图像的曝光值。
  14. 如权利要求12所述的多曝光图像融合方法，其中，所述步骤S2中，权重计算公式为：
    Figure PCTCN2019070915-appb-100006
    其中，W k(i,j)为第K张曝光图像的第i行第j列像素的权重，L med,k为第K张曝光图像需增强区域亮度均值。
  15. 如权利要求14所述的多曝光图像融合方法，其中，所述步骤S3中，判断原始图像的累积直方图的分布类型的方法为：根据原始图像的累积直方图生成对应原始图像的所有像素的K-1个的概率分布区块，且每一个概率分布区块的最大灰阶与最小灰阶之差相等；
    当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中的多个相邻的概率分布区块的累积概率之和大于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为中间分布型；
    当K-1个相等的概率分布区块中除第1个概率分布区块与第K-1个概率分布区块之外的其中多个相邻的概率分布区块的累积概率之和小于或等于一预设的累积概率之和，判断原始图像的累积直方图的分布类型为两端分布型。
PCT/CN2019/070915 2018-11-30 2019-01-08 多曝光图像融合方法 WO2020107662A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811460274.3A CN109636767A (zh) 2018-11-30 2018-11-30 多曝光图像融合方法
CN201811460274.3 2018-11-30

Publications (1)

Publication Number Publication Date
WO2020107662A1 true WO2020107662A1 (zh) 2020-06-04

Family

ID=66070553

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/070915 WO2020107662A1 (zh) 2018-11-30 2019-01-08 多曝光图像融合方法

Country Status (2)

Country Link
CN (1) CN109636767A (zh)
WO (1) WO2020107662A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP4283556A4 (en) * 2021-01-29 2023-11-29 Beijing Bytedance Network Technology Co., Ltd. IMAGE PROCESSING METHOD AND APPARATUS, ELECTRONIC DEVICE AND MEDIUM

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110087003B (zh) * 2019-04-30 2021-03-23 Tcl华星光电技术有限公司 多曝光图像融合方法
CN110211077B (zh) * 2019-05-13 2021-03-09 杭州电子科技大学上虞科学与工程研究院有限公司 一种基于高阶奇异值分解的多曝光图像融合方法
CN110619610B (zh) * 2019-09-12 2023-01-10 紫光展讯通信(惠州)有限公司 图像处理方法及装置
CN110728647A (zh) * 2019-10-16 2020-01-24 紫光展讯通信(惠州)有限公司 水下图像的增强方法及装置、存储介质、终端
CN110719407A (zh) * 2019-10-18 2020-01-21 北京字节跳动网络技术有限公司 图片美化方法、装置、设备及存储介质
CN114187213A (zh) * 2021-12-14 2022-03-15 成都微光集电科技有限公司 图像融合方法及其装置、设备和存储介质
EP4343693A1 (en) * 2022-01-10 2024-03-27 Honor Device Co., Ltd. Image processing method and related device

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4600684B2 (ja) * 2006-08-18 2010-12-15 日本ビクター株式会社 撮影装置及び撮影方法
CN106408518A (zh) * 2015-07-30 2017-02-15 展讯通信（上海）有限公司 图像融合方法、装置及终端设备
CN106506983A (zh) * 2016-12-12 2017-03-15 天津大学 一种适用于ldr视频的hdr视频生成方法
CN106530263A (zh) * 2016-10-19 2017-03-22 天津大学 一种适应于医学影像的单曝光高动态范围图像生成方法
CN107194884A (zh) * 2017-04-18 2017-09-22 天津大学 一种具有不同曝光值的多幅ldr图像获取方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106454144B (zh) * 2016-09-08 2019-04-02 清华大学 一种对谷歌眼镜图像过度曝光的校正方法
US10706512B2 (en) * 2017-03-07 2020-07-07 Adobe Inc. Preserving color in image brightness adjustment for exposure fusion
CN106920221B (zh) * 2017-03-10 2019-03-26 重庆邮电大学 兼顾亮度分布和细节呈现的曝光融合方法
CN107590791B (zh) * 2017-09-22 2020-07-03 深圳市华星光电半导体显示技术有限公司 图像增强方法以及图像处理装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4600684B2 (ja) * 2006-08-18 2010-12-15 日本ビクター株式会社 撮影装置及び撮影方法
CN106408518A (zh) * 2015-07-30 2017-02-15 展讯通信（上海）有限公司 图像融合方法、装置及终端设备
CN106530263A (zh) * 2016-10-19 2017-03-22 天津大学 一种适应于医学影像的单曝光高动态范围图像生成方法
CN106506983A (zh) * 2016-12-12 2017-03-15 天津大学 一种适用于ldr视频的hdr视频生成方法
CN107194884A (zh) * 2017-04-18 2017-09-22 天津大学 一种具有不同曝光值的多幅ldr图像获取方法

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
ZHENGFANG ET AL: "Direct Fusion Algorithm for Multi-exposed Images Based on Sigmoid Function Fitting", CHINESE JOURNAL OF SCIENTIFIC INSTRUMENT, vol. 36, no. 10, 31 October 2015 (2015-10-31), pages 2321 - 2329, XP009521413, ISSN: 0254-3087, DOI: 10.19650/j.cnki.cjsi.2015.10.021 *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP4283556A4 (en) * 2021-01-29 2023-11-29 Beijing Bytedance Network Technology Co., Ltd. IMAGE PROCESSING METHOD AND APPARATUS, ELECTRONIC DEVICE AND MEDIUM

Also Published As

Publication number Publication date
CN109636767A (zh) 2019-04-16

Similar Documents

Publication Publication Date Title
WO2020107662A1 (zh) 多曝光图像融合方法
CN110087003B (zh) 多曝光图像融合方法
CN108630148B (zh) 显示面板亮度差异的补偿方法及显示器
CN108428436B (zh) 亮度补偿方法、亮度补偿装置、显示装置及存储介质
CN109064979B (zh) 图像显示处理方法及装置、显示装置及存储介质
CN110610463A (zh) 一种图像增强方法及装置
CN104699438B (zh) 对oled显示器的待显示画面进行处理的设备和方法
WO2017143635A1 (zh) 一种降低显示面板功耗的方法及系统
US10332436B2 (en) Luminance adjustment system
CN109147685B (zh) Mura补偿方法
WO2019165830A1 (zh) 用于显示面板的光学补偿方法和光学补偿设备
US20180247609A1 (en) Locally Adaptive Backlight Control
WO2019100675A1 (zh) 一种分区背光控制方法及其装置
WO2018188122A1 (zh) 一种显示器的驱动方法和驱动装置、显示器
CN113724652B (zh) OLED显示面板Mura的补偿方法、装置及可读介质
CN111816121A (zh) 显示面板的亮度补偿方法、系统及显示面板
WO2019205751A1 (zh) 图像增强方法
CN104751807A (zh) 一种背光亮度调节方法、装置及液晶显示装置
US7956837B2 (en) Method for eliminating shadow around support pin of LED backlight
US20100020003A1 (en) Methods and Systems for Area Adaptive Backlight Management
CN110728944B (zh) 显示装置及显示方法
WO2022126337A1 (zh) 伽马校正方法及装置、电子设备和可读存储介质
CN101772954B (zh) 用于视频回放的管理技术
CN107342054A (zh) 显示装置、显示控制方法及显示控制装置
CN109658864B (zh) 一种显示面板显示处理方法和显示处理装置

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: 19891141

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: 19891141

Country of ref document: EP

Kind code of ref document: A1