Fuzzy image processing method and device and terminal equipment
A technology of blurring images and processing methods, which is applied in the field of image processing and can solve problems such as difficult to effectively eliminate image blurring
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
Image
Examples
Embodiment 1
[0038] figure 1 It shows a flow chart of a blurred image processing method provided by the embodiment of the present application, which is described in detail as follows:
[0039] Step S11, obtaining the generation network and the discrimination network.
[0040] Among them, the generative network is a network that generates images. Specifically, the obtained generation network can be established based on a residual network (Residual Network, ResNet). In order to facilitate optimization and convergence, the ResNet has at least 2 residual blocks. Preferably, the number of residual blocks of the ResNet is 3.
[0041] Among them, the discriminative network is used to judge whether an image is "real". Its input parameter is x, and x represents an image. Assuming that the discriminant network is represented by "D", the output D(x) represents the probability that x is a real image. If the output of the discriminant network is 1, it means that x is a real image. The probability of...
Embodiment 2
[0053] figure 2 It shows a flow chart of another blurred image processing method provided by the embodiment of the present application, which is described in detail as follows:
[0054] Step S21, obtaining the generation network and the discrimination network;
[0055] The step S21 is the same as the step S11 in the first embodiment, and will not be repeated here.
[0056] Step S22, inputting the fuzzy image of the preset data set into the generating network to obtain the generated image output by the generating network;
[0057] In this step, a plurality of fuzzy images in the preset data set are input into the generation network one by one to obtain a generated image output by the generation network.
[0058] Step S23, inputting the generated image into the discrimination network, and establishing a reconstruction cost function of the generation network according to an output result of the discrimination network.
[0059] In this step, the cross-entropy loss function of ...
Embodiment 3
[0088] image 3 It shows a schematic structural diagram of a blurred image processing device provided by the embodiment of the present application. For the convenience of description, only the parts related to the embodiment of the present application are shown:
[0089] The blurred image processing device includes: a network acquisition unit 31 , a reconstruction cost function establishment unit 32 , an adversarial cost function establishment unit 33 , a generation network and discrimination network training unit 34 , and an image processing unit 35 . in:
[0090] A network acquisition unit 31, configured to acquire a generation network and a discrimination network;
[0091] Optionally, the obtained generation network can be established based on ResNet. For the convenience of optimization and convergence, the ResNet has at least two residual blocks. Preferably, the number of residual blocks of the ResNet is three.
[0092] Optionally, the discriminant network can be establi...
PUM
Login to View More Abstract
Description
Claims
Application Information
Login to View More