Shadow detection method assisted by data set expansion and shadow image classification

Abstract

The invention discloses a shadow detection method assisted by data set expansion and shadow image classification, and belongs to the field of image processing. The method comprises the following steps: 1, based on a generative adversarial network, designing a ShadowGAN network structure for generating a shadow image, and expanding an original data set by using the generated shadow image; 2, adding a shadow classification module into an existing shadow detection network model; and 3, the step 1 and the step 2 are combined, so that the detection accuracy is further improved. The invention provides a data set expansion method for shadow detection and a shadow image classification-assisted shadow detection network model. According to the method, for the shadow image obtained in the natural environment, the deep neural network is utilized, the network structure of the generative adversarial network is designed to expand the data set, and the shadow area in the shadow image is recognized more accurately through the network structure of the shadow detection model assisted by shadow image classification.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a shadow detection method. Background technique [0002] The presence of shadows can interfere with some computer vision tasks, such as object detection, object tracking, and semantic segmentation. On the other hand, the shadow also contains information such as the direction of light, the position of the camera, and the geometric shape of the object. Therefore, image shadow detection is an important step. [0003] Early shadow detection methods were designed based on imaging models or artificially designed features. These methods require high image quality and certain lighting conditions, such as Lambert surfaces and Planck light sources. These methods need to meet specific conditions, and it is difficult to apply to different lighting conditions and more complex environments. [0004] Recently, shadow detection methods based on Convolutional Neural Network (CNN) ha...

AI Technical Summary

Problems solved by technology

In addition, the above-mentioned invention CN201910256619.1 only used shadow images and shadow masks when training the shadow detection sub-network, and did not give full play to the role of shadow-free images for shadow detection

Images