点击次数：

论文类型：期刊论文

发表时间：2016-07-12

发表刊物：CHEMICAL ENGINEERING SCIENCE

收录刊物：SCIE、EI

卷号：148

页面范围：126-139

ISSN号：0009-2509

关键字：Crystal morphology; Shape identification; Imaging analysis; Feature analysis; Inner distance descriptor; Crystal size measurement

摘要：A synthetic image analysis strategy is proposed for in-situ crystal size measurement and shape identification for monitoring crystallization processes, based on using a real-time imaging system. The proposed method consists of image processing, feature analysis, particle sieving, crystal size measurement, and crystal shape identification. Fundamental image features of crystals are selected for efficient classification. In particular, a novel shape feature, referred to as inner distance descriptor, is introduced to quantitatively describe different crystal shapes, which is relatively independent of the crystal size and its geometric direction in an image captured for analysis. Moreover, a pixel equivalent calibration method based on subpixel edge detection and circle fitting is proposed to measure crystal sizes from the captured images. In addition, a kernel function based method is given to deal with nonlinear correlations between multiple features of crystals, facilitating computation efficiency for real-time shape identification. Case study and experimental results from the cooling crystallization of L-glutamic acid demonstrate that the proposed image analysis method can be effectively used for in-situ crystal size measurement and shape identification with good accuracy. (C) 2016 Elsevier Ltd. All rights reserved.