期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2018
卷号：96
期号：13
出版社：Journal of Theoretical and Applied
摘要：Recognition of hand drawn electronics components has application in academic and research in electronics engineering. In this paper, we propose an approach to recognize hand drawn electronic components using histogram of oriented gradient (HOG) features and subsequently Support Vector Machine (SVM) classifier is used to classify the components. The objective is to recognize the hand drawn electronics components. In order to achieve best recognition, we consider hand drawn scanned images and converted them to bi-level image then applied morphological operation to remove discontinuity. Further, the proposed method extracts features of intensity gradient and direction. We trained ten components with large data set each with 200 samples and tested with tenfold cross validation. To establish the efficacy of the proposed method, we have conducted experiment on large dataset of about 2000 images. From experiments it is revealed that the proposed method has yielded 96.9% recognition rate.
关键词：Classifier; Circuits; Hand Written Components; HOG Descriptor
