Title :
Lithium battery swollen detection based on computer vision
Author :
Yinyin Zhan ; Jiwei Deng ; Taihong Wang
Author_Institution :
Key Lab. for Micro-Nano-Optoelectron. Devices of Minist. of Educ., Hunan Univ., Changsha, China
Abstract :
In the process of lithium battery production, swollen battery detection mostly depends on manual testing which is subjective and inefficient. A new method based on computer vision is proposed to detect and separate the swollen battery. The mainstream shapes of lithium batteries are square and cylinder. This work uses square shape battery as example. The work first use an improved bimodal histogram method to segment target area, and extract battery geometrical characteristics as feature vectors, then build the classification model using C-SVM. Genetic algorithm is adopted to optimize SVM parameters and k-fold validation strategy is used to evaluate it. Experiment shows that, the proposed method can achieve a recognition rate of 98.1481%, which provides an intelligent and efficient detection method for swollen battery automated separation in the production.
Keywords :
computer vision; electrical engineering computing; genetic algorithms; image classification; image segmentation; secondary cells; support vector machines; C-SVM; Li; battery geometrical characteristics; bimodal histogram method; classification model; computer vision; feature vector extraction; genetic algorithm; k-fold validation strategy; lithium battery production process; lithium battery swollen detection; manual testing; square shape battery; swollen battery automated separation; target area segmentation; Computers; Gray-scale; Histograms; Manuals; lithium battery; machine vision; support vector machine; swollen detection;
Conference_Titel :
Software Engineering and Service Science (ICSESS), 2013 4th IEEE International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4673-4997-0
DOI :
10.1109/ICSESS.2013.6615409