Image quality guided biology application for genetic analysis

Abstract

Image quality assessment (IQA) is widely applied in image processing, such as image retrieval, image aesthetic evaluation and image classification. To expand application of IQA, we propose a novel method toward biology application using IQA for genetic research. The proposed approach breaks through limitations of traditional biology research, which integrates image processing algorithms with biology applications. Specifically, we first conduct the dataset acquisition including frozen semen images and their according biology scores that reflect genetic attributes. Then, each obtained image will be assigned a quality score according to its grayscale features and texture features. Afterward, we leverage BP neural network for deep feature extraction with fusing quality score and biology score as tags. Finally, given a test image, we can predict its genetic attribute according to deep-learned model. Comprehensive experiments conducted on goat genetic research demonstrate satisfied performance of proposed method.