CN110852947A - A super-resolution method for infrared images based on edge sharpening - Google Patents
A super-resolution method for infrared images based on edge sharpening Download PDFInfo
- Publication number
- CN110852947A CN110852947A CN201911046112.XA CN201911046112A CN110852947A CN 110852947 A CN110852947 A CN 110852947A CN 201911046112 A CN201911046112 A CN 201911046112A CN 110852947 A CN110852947 A CN 110852947A
- Authority
- CN
- China
- Prior art keywords
- image
- infrared
- super
- resolution
- infrared image
- 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
Links
- 238000000034 method Methods 0.000 title claims abstract description 18
- 238000012545 processing Methods 0.000 claims abstract description 30
- 238000013528 artificial neural network Methods 0.000 claims abstract description 14
- 238000012549 training Methods 0.000 claims description 23
- 238000000605 extraction Methods 0.000 claims description 4
- 238000011084 recovery Methods 0.000 abstract 1
- 238000001514 detection method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000003384 imaging method Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000003707 image sharpening Methods 0.000 description 1
- 238000003331 infrared imaging Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4046—Scaling of whole images or parts thereof, e.g. expanding or contracting using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/73—Deblurring; Sharpening
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10048—Infrared image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20172—Image enhancement details
- G06T2207/20192—Edge enhancement; Edge preservation
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Image Processing (AREA)
- Studio Devices (AREA)
Abstract
本发明公开了一种基于边缘锐化的红外图像超分辨方法。利用红外摄像机获得红外图像，建立红外超分辨神经网络结构，网络包括图像处理和图像边缘处理两个子网络，针对于输入网络的红外图像，图像处理网络主要用于恢复图像的结构信息，边缘处理网络用于恢复图像的细节边缘信息；其中图像处理网络分为两个阶段，第一阶段实现红外图像去噪并且实现图像的结构信息恢复，第二阶段实现图像超分辨并实现更多的图像细节结构信息恢复。本发明基于数字红外图像超分辨的要求，通过图像结构和边缘信息的分别处理实现了高倍率的红外图像超分辨。
The invention discloses an infrared image super-resolution method based on edge sharpening. Using an infrared camera to obtain infrared images, an infrared super-resolution neural network structure is established. The network includes two sub-networks: image processing and image edge processing. For the infrared image input to the network, the image processing network is mainly used to restore the structural information of the image, and the edge processing network It is used to restore the detailed edge information of the image; the image processing network is divided into two stages, the first stage realizes infrared image denoising and realizes the restoration of image structure information, and the second stage realizes image super-resolution and realizes more image detail structure Information recovery. Based on the requirements of digital infrared image super-resolution, the present invention realizes high-magnification infrared image super-resolution by separately processing the image structure and edge information.
Description
技术领域technical field
本发明属于数字成像技术领域的红外图像超分辨算法，具体涉及一种基于边缘锐化的红外图像超分辨方法。The invention belongs to an infrared image super-resolution algorithm in the technical field of digital imaging, in particular to an infrared image super-resolution method based on edge sharpening.
技术背景technical background
随着红外探测技术的发展，采用红外成像技术进行目标识别，提高对目标的智能探测和检测识别能力，在对红外图像的探测过程中，受到目标的边缘轮廓特征干扰的影响，导致红外图像的输出质量不好，降低了目标单帧红外图像的识别和检测能力。With the development of infrared detection technology, infrared imaging technology is used for target recognition to improve the ability of intelligent detection and detection and recognition of targets. The output quality is not good, which reduces the recognition and detection ability of the target single frame infrared image.
高分辨率图像可以提供比其对应的低分辨率图像更多的细节。这些细节在所有领域都应该是至关重要的。由于硬件设备的局限性，超分辨已广泛应用于许多成像设备。超分辨是在不改变镜头焦距的情况下，把图像进行了数字放大，因此导致了图像质量的下降：然而，图像处理算法(例如图像插值)除锯齿和模糊伪像外不会产生高质量图片。为解决这个问题，过去的几十年中已经提出了许多的改进算法。例如使用插值来增加输入图像的空间分辨率，基于内插的恢复方法旨在搜索相邻像素之间的连接并且逐个填充缺失像素函数或内插核等等。虽然它在低计算复杂度下具有快速处理时间，但是逐步运算的方法不能保证估计的准确性，尤其是在存在噪声的情况下。有些文献提出用神经网络的方法对红外图像进行超分辨，但由于红外图像噪声偏大导致输出图像质量并不高。High-resolution images can provide more detail than their lower-resolution counterparts. These details should be critical in all areas. Due to the limitations of hardware devices, super-resolution has been widely used in many imaging devices. Super-resolution digitally enlarges the image without changing the focal length of the lens, thus resulting in a reduction in image quality: however, image processing algorithms (such as image interpolation) do not produce high-quality images except for aliasing and blurring artifacts . To solve this problem, many improved algorithms have been proposed in the past few decades. For example, using interpolation to increase the spatial resolution of the input image, interpolation-based restoration methods aim to search for connections between adjacent pixels and fill missing pixel functions or interpolation kernels one by one, etc. Although it has fast processing time with low computational complexity, the step-by-step approach cannot guarantee the accuracy of the estimation, especially in the presence of noise. Some literatures propose to use the neural network method to super-resolve infrared images, but the output image quality is not high due to the high noise of infrared images.
发明内容SUMMARY OF THE INVENTION
为了解决背景技术中的问题，本发明提供了一种基于边缘锐化的红外图像超分辨方法，提升了红外图像超分辨的成像质量，针对不同放大倍率图像进行设计，在利用红外图像边缘和锐化进行修复的方式上提出了新的方法。In order to solve the problems in the background technology, the present invention provides an infrared image super-resolution method based on edge sharpening, which improves the imaging quality of infrared image super-resolution, and is designed for images with different magnifications. A new method is proposed in the way of repairing.
本发明基于神经网络，在网络中间进行图像的去噪和图像细节的恢复，通过对图像的边缘进行提取以加强图像的细节信息，并且对图像进行锐化处理使输出的图像细节更加丰富。Based on the neural network, the invention denoises the image and restores the image details in the middle of the network, enhances the detail information of the image by extracting the edge of the image, and sharpens the image to enrich the details of the output image.
本发明采用的技术方案包括以下步骤：The technical scheme adopted in the present invention comprises the following steps:
步骤1：选取红外相机对场景或对象进行拍摄获得的红外图像和其对应的训练目标图像作为训练集；Step 1: Select an infrared image obtained by shooting a scene or an object with an infrared camera and its corresponding training target image as a training set;
步骤2：建立红外图像超分辨神经网络，红外图像超分辨神经网络结构包括图像处理子网络和图像边缘处理子网络；Step 2: establish an infrared image super-resolution neural network, and the infrared image super-resolution neural network structure includes an image processing sub-network and an image edge processing sub-network;
步骤3：将训练集输入红外图像超分辨神经网络进行训练；Step 3: Input the training set into the infrared image super-resolution neural network for training;
步骤4：将用红外相机对场景或对象进行拍摄获得的待修复红外图像输入步骤3训练后的红外图像超分辨神经网络，得到待修复红外图像的超分辨图像。Step 4: Input the infrared image to be repaired obtained by photographing the scene or object with an infrared camera into the infrared image super-resolution neural network trained in step 3 to obtain a super-resolution image of the infrared image to be repaired.
所述步骤2具体为：The step 2 is specifically:
2.1)图像处理子网络包括阶段一和阶段二两个阶段；2.1) The image processing sub-network includes two stages: stage one and stage two;
将红外图像输入阶段一进行多组卷积操作后提取的特征和输入的红外图像逐像素相加，获得通过高斯去噪后的模糊图像并恢复红外图像的低频信息；The features extracted after multiple convolution operations are performed in the infrared image input stage 1 and the input infrared image are added pixel by pixel to obtain the blurred image after Gaussian denoising and restore the low-frequency information of the infrared image;
将阶段一获得的高斯去噪后的图像输入阶段二继续进行多组卷积操作，恢复红外图像的高频信息；The Gaussian denoised image obtained in stage 1 is input into stage 2, and multiple groups of convolution operations are continued to restore the high-frequency information of the infrared image;
2.2)将红外图像输入图像边缘处理子网络进行卷积操作，提取红外图像的边缘细节特征生成边缘图像；2.2) Input the infrared image into the image edge processing sub-network to perform a convolution operation, and extract the edge detail features of the infrared image to generate an edge image;
2.3)将步骤2.2)生成的边缘图像和步骤2.1)阶段二的处理结果进行逐像素相加生成最终的输出图像，即得红外图像的超分辨图像。2.3) Perform pixel-by-pixel addition of the edge image generated in step 2.2) and the processing result of stage 2 in step 2.1) to generate a final output image, that is, a super-resolution image of the infrared image.
所述步骤1中的训练目标图像为：The training target image in step 1 is:
训练目标图像一：对红外图像进行模糊核为7、方差为3的高斯模糊处理以实现图像的去噪，作为步骤2.1)中图像处理子网络阶段一的训练目标图像；Training target image 1: The infrared image is subjected to Gaussian blurring with a blur kernel of 7 and a variance of 3 to achieve image denoising, as the training target image of image processing sub-network stage 1 in step 2.1);
训练目标图像二：对红外图像用sobel算子进行图像的边缘提取，作为步骤2.2)图像边缘处理子网络的训练目标图像；Training target image two: carry out the edge extraction of the image with the sobel operator to the infrared image, as step 2.2) the training target image of the image edge processing sub-network;
训练目标图像三：对红外图像进行锐化处理，作为步骤2.3)的训练目标图像。Training target image 3: Sharpen the infrared image as the training target image in step 2.3).
所述步骤2.3)获得的红外图像的超分辨图像为输入的红外图像尺寸的2倍或4倍，超分辨图像的分辨率高于输入的红外图像。The super-resolution image of the infrared image obtained in the step 2.3) is 2 times or 4 times the size of the input infrared image, and the resolution of the super-resolution image is higher than that of the input infrared image.
本发明的有益效果如下：The beneficial effects of the present invention are as follows:
(1)本发明基于红外图像超分辨的要求，通过图像卷积网络的结构实现了数字图像任意倍率的超分辨。(1) Based on the requirements of infrared image super-resolution, the present invention realizes the super-resolution of digital images at any magnification through the structure of the image convolution network.
(2)本发明第一次在红外图像超分辨的过程中对红外图像进行了去噪处理，并且利用锐化图像和增强图像边缘信息提高了图像的细节，相比于现有的技术，在图像的视觉效果上有明显的提升。(2) For the first time in the present invention, the infrared image is denoised in the process of infrared image super-resolution, and the image details are improved by sharpening the image and enhancing the edge information of the image. The visual effect of the image has been significantly improved.
附图说明Description of drawings
图1是红外图像超分辨神经网络结构示意图；Figure 1 is a schematic diagram of the structure of an infrared image super-resolution neural network;
图2是实施例1进行2倍超分辨的输入图；Fig. 2 is the input graph that embodiment 1 carries out 2 times super-resolution;
图3是实施例1的输入图采用双三次插值、VDSR以及本发明方法的对比结果；Fig. 3 is the comparison result that the input graph of embodiment 1 adopts bicubic interpolation, VDSR and the inventive method;
图4是实施例1的输入图采用双三次插值、VDSR以及本发明方法的细节对比结果图；4 is a detailed comparison result diagram of the input graph of Embodiment 1 using bicubic interpolation, VDSR and the method of the present invention;
图5是示例性实施例2进行4倍超分辨的输入图；FIG. 5 is an input graph for 4-fold super-resolution performed by Exemplary Embodiment 2;
图6是实施例2的输入图采用双三次插值、VDSR以及本发明方法的对比结果图；Fig. 6 is the comparison result diagram that the input graph of embodiment 2 adopts bicubic interpolation, VDSR and the inventive method;
图7是实施例2的输入图采用双三次插值、VDSR以及本发明方法的细节对比结果图。FIG. 7 is a detailed comparison result diagram of the input image of Embodiment 2 using bicubic interpolation, VDSR and the method of the present invention.
具体实施方式Detailed ways
下面结合附图和实施例对本发明做近一步描述。The present invention will be further described below with reference to the accompanying drawings and embodiments.
如图1所示，红外图像超分辨神经网络结构包括图像处理子网络和图像边缘处理子网络；As shown in Figure 1, the infrared image super-resolution neural network structure includes an image processing sub-network and an image edge processing sub-network;
1)图像处理子网络包括阶段一和阶段二两个阶段；1) The image processing sub-network includes two stages: stage one and stage two;
在阶段一，输入的红外图像通过多层卷积层进行图像特征的提取，卷积可以由公式(1)表示：In the first stage, the input infrared image is extracted by the multi-layer convolution layer, and the convolution can be expressed by formula (1):
Fgi＝Convi(I) (1)F gi = Convi (I) (1)
其中，Convi(I)表示步长为i的卷积，Fgi表示阶段一第i个卷积提取出的特征，I表示输入图像或特征；Wherein, Conv i (I) represents the convolution with step i, F gi represents the feature extracted by the i-th convolution in stage 1, and I represents the input image or feature;
将输入的红外图像和阶段一卷积层提取的特征进行逐像素相加获得高斯去噪图像；将阶段一生成的高斯去噪图像输入到阶段二的多个卷积层当中对图形进行更进一步的特征提取。Add the input infrared image and the features extracted by the convolutional layer of the first stage to obtain a Gaussian denoised image; input the Gaussian denoised image generated in the first stage into the multiple convolutional layers of the second stage to further the graph feature extraction.
2)红外图像输入图像边缘处理子网络，直接进行卷积层提取生成边缘图像。2) The infrared image is input to the image edge processing sub-network, and the convolution layer is directly extracted to generate the edge image.
3)最后将边缘图像与阶段二的处理结果进行逐像素相加，生成最终的输出图像。3) Finally, the edge image and the processing result of the second stage are added pixel by pixel to generate the final output image.
在训练阶段将阶段一生成的高斯去噪图像与高斯模糊处理后红外图像的图像进行对比，将图像边缘处理子网络生成的边缘图像与sobel算子提取的图像边缘目标图像进行对比，分别通过计算L1范数作为损失函数。将最后生成的输出图像与锐化后的红外图像进行对比，计算L1范数作为损失函数。将三个损失函数进行加权求和作为整个网络的损失函数，使用Adam优化器对网络参数进行优化In the training stage, the Gaussian denoising image generated in the first stage is compared with the image of the infrared image after Gaussian blurring, and the edge image generated by the image edge processing sub-network is compared with the image edge target image extracted by the sobel operator. L1 norm as loss function. The final generated output image is compared with the sharpened infrared image, and the L1 norm is calculated as the loss function. The weighted summation of the three loss functions is used as the loss function of the entire network, and the network parameters are optimized using the Adam optimizer
本发明的具体实施例如下：Specific embodiments of the present invention are as follows:
本发明具体实施包括阶段一和阶段二两个阶段。在阶段一，将红外图片输入网络，用九个步长为1，尺寸为3*3*64的卷积核以及一个步长为1，尺寸为3*3*1的卷积核对图像进行卷积操作，然后与输入的红外图像进行逐像素相加生成高斯去噪图像；同时用五个步长为1，尺寸为3*3*64的卷积核以及一个步长为1，尺寸为3*3*1的卷积核对图像进行卷积操作生成边缘图像。然后用5个步长为1，尺寸为3*3*64的卷积核以及一个步长为1，尺寸为3*3*1的卷积核对阶段一生成的高斯去噪图像进行卷积，得到最后的输出图像。The specific implementation of the present invention includes two stages: stage one and stage two. In stage 1, the infrared image is input to the network, and the image is rolled with nine convolution kernels with stride 1 and size 3*3*64 and a convolution kernel with stride 1 and size 3*3*1. product operation, and then perform pixel-by-pixel addition with the input infrared image to generate a Gaussian denoised image; at the same time, five convolution kernels with a stride of 1 and a size of 3*3*64 and a stride of 1 and a size of 3 are used. The convolution kernel of *3*1 performs the convolution operation on the image to generate the edge image. Then convolve the Gaussian denoised image generated in stage 1 with 5 convolution kernels with stride 1 and size 3*3*64 and a convolution kernel with stride 1 and size 3*3*1. to get the final output image.
针对不同缩放倍数的分辨率要求，可以对输入图像进行相应倍率的插值操作，从而得到不同倍率的超分辨图像。According to the resolution requirements of different zoom ratios, the input image can be interpolated at the corresponding magnification to obtain super-resolution images of different magnifications.
本发明使用图1所示结构，分别对图2、图5所示的红外图像进行2倍分辨率和4倍分辨率成像，并与双三次插值以及VDSR算法对比，说明本发明的有益效果。The present invention uses the structure shown in FIG. 1 to image the infrared images shown in FIG. 2 and FIG. 5 with 2 times resolution and 4 times resolution respectively, and compares with bicubic interpolation and VDSR algorithm to illustrate the beneficial effects of the present invention.
如图3～图4、图6～图7所示，对比本发明的方法与双三次插值和VDSR所产生的超分辨图像，不管是2倍分辨率还是4倍分辨率图像，都可以发现本发明的方法生成图像的纹理更加丰富，细节更加明显。As shown in Figures 3 to 4 and Figures 6 to 7, comparing the method of the present invention with the super-resolution images generated by bicubic interpolation and VDSR, whether it is a 2x resolution or a 4x resolution image, it can be found that the present invention The inventive method generates images with richer textures and more obvious details.
在红外图像超分辨系统中，为了重建高分辨率的红外图像，首次提出用边缘提取和图片锐化的损失函数结构的神经网络超分辨算法。同时，将结合了高斯去噪图像损失已达到图像去噪的目的，本发明显著改善了红外图像超分辨的成像质量。In the infrared image super-resolution system, in order to reconstruct high-resolution infrared images, a neural network super-resolution algorithm with a loss function structure of edge extraction and image sharpening is proposed for the first time. At the same time, the Gaussian denoising image loss has been combined to achieve the purpose of image denoising, and the present invention significantly improves the imaging quality of infrared image super-resolution.
Claims (4)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911046112.XA CN110852947B (en) | 2019-10-30 | 2019-10-30 | A super-resolution method for infrared images based on edge sharpening |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911046112.XA CN110852947B (en) | 2019-10-30 | 2019-10-30 | A super-resolution method for infrared images based on edge sharpening |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110852947A true CN110852947A (en) | 2020-02-28 |
| CN110852947B CN110852947B (en) | 2021-07-20 |
Family
ID=69599307
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911046112.XA Active CN110852947B (en) | 2019-10-30 | 2019-10-30 | A super-resolution method for infrared images based on edge sharpening |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110852947B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932460A (en) * | 2020-08-10 | 2020-11-13 | 北京大学深圳医院 | MR image super-resolution reconstruction method and device, computer equipment and storage medium |
| CN113674176A (en) * | 2021-08-23 | 2021-11-19 | 北京市商汤科技开发有限公司 | Image restoration method and device, electronic equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106897986A (en) * | 2017-01-23 | 2017-06-27 | 浙江大学 | A kind of visible images based on multiscale analysis and far infrared image interfusion method |
| CN107392852A (en) * | 2017-07-10 | 2017-11-24 | 深圳大学 | Super resolution ratio reconstruction method, device, equipment and the storage medium of depth image |
| WO2017219263A1 (en) * | 2016-06-22 | 2017-12-28 | 中国科学院自动化研究所 | Image super-resolution enhancement method based on bidirectional recursion convolution neural network |
| CN108961180A (en) * | 2018-06-22 | 2018-12-07 | 理光软件研究所（北京）有限公司 | infrared image enhancing method and system |
| CN109389552A (en) * | 2017-08-02 | 2019-02-26 | 中山大学 | A kind of Image Super-resolution based on context-sensitive multitask deep learning |
| CN109801215A (en) * | 2018-12-12 | 2019-05-24 | 天津津航技术物理研究所 | The infrared super-resolution imaging method of network is generated based on confrontation |
-
2019
- 2019-10-30 CN CN201911046112.XA patent/CN110852947B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017219263A1 (en) * | 2016-06-22 | 2017-12-28 | 中国科学院自动化研究所 | Image super-resolution enhancement method based on bidirectional recursion convolution neural network |
| CN106897986A (en) * | 2017-01-23 | 2017-06-27 | 浙江大学 | A kind of visible images based on multiscale analysis and far infrared image interfusion method |
| CN107392852A (en) * | 2017-07-10 | 2017-11-24 | 深圳大学 | Super resolution ratio reconstruction method, device, equipment and the storage medium of depth image |
| CN109389552A (en) * | 2017-08-02 | 2019-02-26 | 中山大学 | A kind of Image Super-resolution based on context-sensitive multitask deep learning |
| CN108961180A (en) * | 2018-06-22 | 2018-12-07 | 理光软件研究所（北京）有限公司 | infrared image enhancing method and system |
| CN109801215A (en) * | 2018-12-12 | 2019-05-24 | 天津津航技术物理研究所 | The infrared super-resolution imaging method of network is generated based on confrontation |
Non-Patent Citations (3)
| Title |
|---|
| YI LIU等: "Low Resolution Cell Image Edge Segmentation Based on Convolutional Neural Network", 《IEEE XPLORE》 * |
| 管志强等: "基于模糊集的自适应红外图像边缘锐化算法", 《光子学报》 * |
| 邵保泰等: "基于生成对抗网络的单帧红外图像超分辨算法", 《红外与毫米波学报》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111932460A (en) * | 2020-08-10 | 2020-11-13 | 北京大学深圳医院 | MR image super-resolution reconstruction method and device, computer equipment and storage medium |
| CN111932460B (en) * | 2020-08-10 | 2023-09-22 | 北京大学深圳医院 | MR image super-resolution reconstruction method, device, computer equipment and storage medium |
| CN113674176A (en) * | 2021-08-23 | 2021-11-19 | 北京市商汤科技开发有限公司 | Image restoration method and device, electronic equipment and storage medium |
| CN113674176B (en) * | 2021-08-23 | 2024-04-16 | 北京市商汤科技开发有限公司 | Image restoration method and device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110852947B (en) | 2021-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Anwar et al. | Densely residual laplacian super-resolution | |
| Ignatov et al. | Replacing mobile camera isp with a single deep learning model | |
| CN109102462B (en) | Video super-resolution reconstruction method based on deep learning | |
| Engin et al. | Cycle-dehaze: Enhanced cyclegan for single image dehazing | |
| WO2020015167A1 (en) | Image super-resolution and non-uniform blur removal method based on fusion network | |
| CN111091503A (en) | Image defocus blur method based on deep learning | |
| Xu et al. | Attentive deep network for blind motion deblurring on dynamic scenes | |
| Lee et al. | NTIRE 2025 challenge on efficient burst hdr and restoration: Datasets, methods, and results | |
| Yang et al. | Crnet: A detail-preserving network for unified image restoration and enhancement task | |
| Zhan et al. | Video deblurring via motion compensation and adaptive information fusion | |
| Zhang et al. | Deep motion blur removal using noisy/blurry image pairs | |
| Xiao et al. | Degradation model learning for real-world single image super-resolution | |
| Poonia et al. | An Improved Image Up-Scaling Technique using Optimize Filter and Iterative Gradient Method | |
| CN113096032B (en) | Non-uniform blurring removal method based on image region division | |
| Chen et al. | Attention-based broad self-guided network for low-light image enhancement | |
| Liu et al. | Siamtrans: zero-shot multi-frame image restoration with pre-trained siamese transformers | |
| Chang et al. | Beyond Camera Motion Blur Removing: How to Handle Outliers in Deblurring. | |
| CN110852947B (en) | A super-resolution method for infrared images based on edge sharpening | |
| Chen et al. | Missing recovery: Single image reflection removal based on auxiliary prior learning | |
| Holla et al. | EFID: edge-focused image denoising using a convolutional neural network | |
| Chi et al. | Joint demosaicking and blind deblurring using deep convolutional neural network | |
| Ollion et al. | Joint self-supervised blind denoising and noise estimation | |
| Zhu et al. | LLISP: Low-light image signal processing net via two-stage network | |
| Lal et al. | A comparative study on CNN based low-light image enhancement | |
| Ko et al. | Deep Degradation Prior for Real-World Super-Resolution. |
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 |