Convolutional neural network defogging algorithm based on regional division and heavy fog preprocessing

TL;DR: In this paper, an endoscope image defogging system based on artificial intelligence is presented, which consists of an endoscopy, a cold light source, a cutter, a controller, and an endoscopic image processor.

TL;DR: In this paper, a system and a fast response access control method based on defogging artificial intelligence (AI) image analysis is provided, which realizes the scene mode judgment of the current image through the AI learning technology, an image local area division strategy adapted to scene mode is adopted, local scene depth differences are reduced, the local region division strategy with optimal adaptation to the image scene minimizes the error caused by the assumed local area scene transmissivity equalization, significantly reducing the blocking effect of the defogging image.

TL;DR: In this paper, a neural network for image restoration and a training and using method thereof are disclosed, the neural network comprises at least one subnetwork, the sub-network is provided with a priori embeddinglayer (1), the prior-a embedding layer is used for extracting a specific channel feature from input information input into the prior a prior-b layer and obtaining a first feature map from the input information.

TL;DR: In this paper, a smoke detection application-oriented preprocessing method was proposed, which comprises three steps of: pre-classifying input images into foggy images, light-fog images, and fogless images.

TL;DR: In this paper, an intelligent unmanned surface vessel visual system video rapid sea fog removal method is presented. But the method of the present method is limited to the video image enhancement field, in particular relates to an intelligent UAV visual system.

TL;DR: In this article, a single image defogging method was proposed based on a back propagation neural network (BP neural network) for obtaining fogless image information and calculating to obtain a corresponding fog image; by using RGB (Red Green Blue) values of pixel points of the fog image as inputs and using scene depths of the pixel points from the image as outputs, constructing a BP neural network model and generating a mapping relation between the pixel values and the depth values of fog image in the trained BP neural networks.

TL;DR: In this article, a multi-scale extraction of texture structure characteristics of a foggy image is performed via a sparse autoencoder, and at the same time, various color characteristics related with fog are extracted.

TL;DR: In this paper, a method for removing fog in an image capable of providing a defogged image without a sense of difference by adaptively controlling defogging parameters in response to a density of fog to remove fog.