文章基本信息

标题：ANALYSIS AND PERFORMANCE OF BACK PROPAGATION NEURAL NETWORK FOR IDENTIFICATION RANGE OF MOTION USING BACKGROUND SUBTRACTION AND FEATURE EXTRACTION
本地全文：下载
作者：FAUZIAH ; EP WIBOWO ; HUSTINAWATY 等
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：2019
卷号：97
期号：3
页码：957-967
出版社：Journal of Theoretical and Applied
摘要：The current image processing is used to identify, classification of other process related to the two, can represent the characteristics of creature image optimal for the analysis by algorithm that enable large computing power for it also a combination of computational processing in the image and related computer algorithm or methods used. The human being is the field done by adapting the ability of the method to take information. In the research to get an analysis and performance of hybrid method with Artificial Neural Network for identification of motion the human array the Range of Motion (ROM) in Coronal Plane, the extract is done by background method on binary image. Identification human movement, in each frame extraction results then performed image processing to detect type of the human body motion. Image segmentation is done using background subtraction method to get result into binary image with morphological operation against binary image in the form filling holes and opening area to eliminated noise and improve the result of segmentation, skeleton process and perform extraction. Feature extracted is used to distinguish motion between body object one with the other object body is done through the analysis of morphological forms based on seven grades of moment invariant previously been extracted last step can performed to identifying using Neural Network and to produce the great accuracy and exactly the Range of Motion with 1000 epoch for training. Being in BPNN by combining Moment Invariant for feature extraction whose performance is the highest accuracy about 97.8666%.
关键词：Back Propagation; Background Subtraction; Range of Motion