دانلود رایگان مقاله توصیف اشیاء با ویژگی سیکا برای طبقه بندی چندسطحی

عنوان فارسی
توصیف اشیاء با ویژگی های سیکا برای طبقه بندی چندسطحی
عنوان انگلیسی
Characterizing objects with SIKA features for multiclass classification
صفحات مقاله فارسی
0
صفحات مقاله انگلیسی
11
سال انتشار
2015
نشریه
الزویر - Elsevier
فرمت مقاله انگلیسی
PDF
کد محصول
E322
رشته های مرتبط با این مقاله
مهندسی برق و مهندسی کامپیوتر
گرایش های مرتبط با این مقاله
مهندسی نرم افزار و مهندسی الکترونیک
مجله
محاسبات نرم کاربردی - Applied Soft Computing
دانشگاه
دانشکده الکترونیک و علوم کامپیوتر، دانشگاه علوم و فناوری هواری بومدین، الجزایر
کلمات کلیدی
طبقه بندی شی، SIKA، ماشین پیچیدگی حداقل، ماشین بردار پشتیبان
۰.۰ (بدون امتیاز)
امتیاز دهید
چکیده

Abstract


This paper presents a novel approach for multiclass classification by fusion of KAZE and Scale Invariant Feature Transform (SIFT) features followed by Minimal Complexity Machine (MCM) as the classifier. Unlike the existing features, the paper proposes a new feature SIKA to represent characteristics of an object, as opposed to just forming a compendium of interest points in an image to represent the object characteristics. Further we append a strong and lightweight classifier, MCM to the technique. The resulting classifier easily outperforms existing techniques based on handcrafted features. Two new scores Keypoint Overlap Score (KOS) and Mean Keypoint Overlap Score (MKOS) have also been proposed as part of this work which are useful in establishing the strength of features for object representation.

نتیجه گیری

8. Conclusion


This paper proposed a Fuzzy MIN-MAX combination algorithm, as a strategy to improve the writer’s soft-biometrics prediction. First, three SVM predictors associated to different data features were developed to perform writer’s gender, handedness and age range prediction. Thereafter, SVM responses are combined to improve the prediction accuracy. Comprehensive experiments using two English and Arabic handwritten text datasets, demonstrated that the proposed combination algorithm can considerably improve the prediction accuracy. Also, what we could observe is that the comparison to various combination rules as well as the state of the art, confirmed the effectiveness of this approach. Based on the results reported in this work, we believe that efficiency of the FuzzyMIN-MAXcombinationcanbedemonstratedbetter when using larger datasets. It is important to note that the combination process, handles SVM outputs provided with relevance information. This property allows the use of any kind of data features and makes it reproducible to any other pattern recognition task. To improve again our results, we intend in a future work, to investigate more robust features such as Histogram of Templates as well as new classifiers such as Artificial Immune Recognition Systems (AIRS).


بدون دیدگاه