.png){kind=icon}
返回
View adjustment: helping users improve photographic composition
.png){kind=icon}
PRE
AI
DOI：10.1007/s00530-024-01490-x.png)
.png){kind=icon}
原文求助.png){kind=icon}
分享.png){kind=icon}
收藏
摘要
En 中文
Image composition has always been one of the primary considerations for professional photography, and good composition can greatly enhance the perception of an image. However, it is difficult for most ordinary users to compose images quickly. To solve this challenge, we propose a new view adjustment method to improve image composition in this paper, assisting users in adjusting views promptly. We establish a composition feature extraction module, which can extract composition information of images based on professional photographic composition rules. Based on this module, we construct an image aesthetic assessment network to score the adjustment results. Additionally, we employ the concept of the hyper-net in our design to build a parameter generation-prediction network, enabling the application of view adjustment rules. In addition, we create a usable view adjustment dataset based on an existing image cropping dataset, which includes samples of displacement adjustment, scale adjustment, rotation adjustment, and their combined adjustment types to facilitate model training. Through experimentation, our model has demonstrated outstanding performance with an MAE score of 0.0392, an MSE score of 0.0037, and an IoU score of 0.783. Experimental results indicate that our proposed view adjustment method effectively improves the image composition, enabling real-time guidance for users to capture well-composed photographs.
Keyword:
Image composition
View adjustment
Parameter generation-prediction network
Composition feature
期刊
IF:
3.1
论文数:
2.8K
被引数:
2.7K
机构
引用论文
Cost and performance data for residential buildings fitted with GSHP systems in Melbourne Australia
OAAI
OAAI分享收藏Population-based screening in a municipality after a primary school outbreak of the SARS-CoV-2 Alpha variant, the Netherlands, December 2020–February 2021在荷兰，2020年12月至2021年2月期间，针对某市在初级学校暴发SARS-CoV-2 Alpha变异株后的基于人群的筛查.png)
PLOS ONE
OAAI
PLOS ONEIF0
OAAI分享收藏Do Perceptions of Competence Mediate The Relationship Between Fundamental Motor Skill Proficiency and Physical Activity Levels of Children in Kindergarten?能力的感知是否可以介导幼儿园儿童的基本运动技能熟练程度与身体活动水平之间的关系？
PREAI
分享收藏
查看更多内容