Research on Target Extraction Technology of Fruit and Vegetable Images in the Complex Environment

• 作者: Hao Zheng Yanqing Wang
• 来源期刊: 国际计算机前沿大会会议论文集
• 学科: 社会科学
• 关键词: MANIFOLD RANKING SIGNIFICANCE IMAGES GRADIENT IMAGES Image SEGMENT
• 年，卷（期）: 2017,（1）
• 页码范围: 175-176
• 页数: 2页
• 分类号: C5

摘要

Considering the difficulty of fruit and vegetable images with uneven illumination and uncontrolled backgrounds,this paper proposed an image preprocess algorithm based on visual subject detection.Firstly,we can use manifold ranking to significance test and to acquire significance images,then use gradient images and position weighting to get cutting images,finally adjust the image brightness and remove image noise to complete the image preprocessing.The experiment results demonstrated the robustness and real-time of the proposed algorithm which can segment images accurately and the accuracy is higher than 91%.