Laser & Optoelectronics Progress, Volume. 59, Issue 4, 0410002(2022)
Automatic Background Blurring Algorithm Based on Image Perception and Segmentation
Aiming at the problem that it is difficult for monocular mobile devices to capture background blurring images, this paper proposes an automatic background blurring algorithm based on the image perception and segmentation algorithm of deep learning. We use the image perception and segmentation algorithm of deep learning to obtain three auxiliary images, i.e., the focus map, depth map, and mask map, of the captured image. We use the auxiliary images to automatically determine the subject or specify the subject by the user, and the depth of each area of the background is calculated. Then, the multi-scale Gaussian filter is used to blur each area of the background in different degrees. Finally, the blurred background areas are merged with the subject, and the edges are optimized to finally generate a background blurred image. Experimental results show that this algorithm can realize more accurate and flexible image virtual processing by using perceptual map based on deep learning, and can automatically focus the image subject or refocus the designated area in a variety of scenes. The virtual effect is natural and hierarchical, and can better highlight the image theme.
Get Citation
Copy Citation Text
Chengmin Liu, Wujian Ye, Yijun Liu. Automatic Background Blurring Algorithm Based on Image Perception and Segmentation[J]. Laser & Optoelectronics Progress, 2022, 59(4): 0410002
Category: Image Processing
Received: Jan. 25, 2021
Accepted: Mar. 19, 2021
Published Online: Jan. 25, 2022
The Author Email: Ye Wujian (yewjian@126.com)