A Fast and Efficient Haar-Like Feature Selection Algorithm for Object Detection

Chung, Byung Woo; Park, Ki-Yeong; Hwang, Sung Yeon

doi:10.7840/kics.2013.38a.6.486

Cited by 9 publications

(3 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Ide dari Haar-like features adalah untuk mengenali obyek berdasarkan nilai sederhana dari fitur tetapi bukan merupakan nilai piksel dari image obyek tersebut. Metode ini memiliki kelebihan yaitu komputasinya sangat cepat, karena hanya bergantung pada jumlah piksel dalam persegi bukan setiap nilai piksel dari sebuah citra (Chung, 2013). Pada penelitian ini Metode Haar-like features digunakan untuk mendeteksi mata dan mulut pada citra wajah seperti yang terlihat pada Gambar 1.…”

Section: Haar-like Featureunclassified

Pengenalan Wajah Menggunakan Implementasi T-shape Mask pada Two Dimentional Linear Discriminant Analysis dan Support Vector Machine

Musthafa

Fitrawan

Supria

2016

JBI

View full text Add to dashboard Cite

Abstract. Face recognition is the identification process to recognize a person's face. Many studies have been developing face recognition methods, one of which is the Two Dimensional Linear Discriminant Analysis (TDLDA) which has pretty good accuracy results with the method of classification Support Vector Machine (SVM). With more training data can add computational time. TDLDA using all the piksel image as input to be processed for feature extraction. Though not all the objects in the area of the face is a significant feature in face recognition. In this study, the proposed use of the T-shape with only use a significant part is the eyes, nose, and mouth are integrated with TDLDA and SVM. The result could reduce computing time on face recognition 21.56% faster than TDLDA method. The accuracy of the results in this study was 91% -96% which is close to the level of accuracy without using a mask on the face.Keyword: face recognition, T-shape, TDLDA, Support vector machine.Â Abstrak. Pengenalan wajah merupakan proses identifikasi untuk mengenali wajah seseorang. Telah Banyak penelitian yang mengembangkan metode pengenalan wajah, salah satunya adalah Two Dimensional Linear Discriminant Analysis (TDLDA) yang memiliki hasil akurasi yang cukup baik dengan metode klasifikasi Support Vector Machine (SVM). Dengan semakin banyak data training dapat menambah waktu komputasinya. TDLDA menggunakan semua piksel citra sebagai masukan yang akan diproses untuk ekstrasi fitur. Padahal tidak semua objek pada area wajah merupakan fitur yang signifikan dalam pengenalan wajah. Dalam penelitian ini diusulkan penggunaan T-shape dengan hanya menyimpan bagian yang signifikan yaitu mata, hidung, dan mulut yang diintegrasikan dengan TDLDA dan SVM. Hasilnya dapat mengurangi waktu komputasi pada pengenalan wajah 21,56% lebih cepat daripada metode TDLDA. Hasil akurasi pada penelitian ini adalah 91%-96% yang mendekati tingkat akurasi tanpa menggunakan mask pada wajah.Kata Kunci: pengenalan wajah, T-shape, TDLDA, Support vector machine.

show abstract

Section: Haar-like Featureunclassified

Pengenalan Wajah Menggunakan Implementasi T-shape Mask pada Two Dimentional Linear Discriminant Analysis dan Support Vector Machine

Musthafa

Fitrawan

Supria

2016

JBI

View full text Add to dashboard Cite

show abstract

“…Tend to receive much influenced by external factors such as changes in the nature of LBP [1] lighting, facial expressions, and backgrounds used the LGP to compensate [2]. Using the CNN algorithm is the method using the S-LGP and U-LGP improved with a Symmetry Uniform and using Cascade Classifier Training [3] method for detecting skin color in the YCbCr space [4] [5] studied an algorithm that can improve the accuracy more [6] [7]. Using this study was an algorithm that can improve the accuracy.…”

Section: Introductionmentioning

confidence: 99%

A Study of Face Detection Using CNN and Cascade Based on Symmetry-LGP & Uniform-LGP and the Skin Color

Kong¹,

Kang

2016

Advanced Science and Technology Letters

View full text Add to dashboard Cite

Abstract. This paper offers toughness to external factors such as lighting, background, angle and improved algorithm on detecting speed. this paper uses the S-LGP(Symmetry Uniform Local Gradient Pattern) and U-LGP(Uniform Local Gradient Pattern) methods that improve the symmetry and uniform using the LGP(Local Gradient Pattern) that are stronger against external factors such as changes of lighting, facial expression, background etc. than the LBP(Local Binary Pattern) through preprocessing the input image and the skin color extracting method at the YCbCr space using the Cascade Classifier Training extractor that forms XML files which is made from gathering and studying the positive face sample and negative face sample at a 1:2 rate and uses the CNN(Convolution Neural Network) of deep learning algorithm to study it. Suggested algorithm is applicable for applications, as well as improve the accuracy of detection.

show abstract

“…AdaBoost was successfully used with Haar-like features in a "cascade" for face detection [5,6] . Using the cascade framework, the background region can be excluded to a great extent from further training.…”

Section: ⅰ Introductionmentioning

confidence: 99%

Real-Time Vehicle License Plate Detection Based on Background Subtraction and Cascade of Boosted Classifiers

Sarker¹,

Song

2014

The Journal of Korean Institute of Communications and Informati

View full text Add to dashboard Cite

License plate (LP) detection is the most imperative part of an automatic LP recognition (LPR) system. Typical LPR contains two steps, namely LP detection (LPD) and character recognition. In this paper, we propose an efficient Vehicle-to-LP detection framework which combines with an adaptive GMM (Gaussian Mixture Model) and a cascade of boosted classifiers to make a faster vehicle LP detector. To develop a background model by using a GMM is possible in the circumstance of a fixed camera and extracts the motions using background subtraction. Firstly, an adaptive GMM is used to find the region of interest (ROI) on which motion detectors are running to detect the vehicle area as blobs ROIs. Secondly, a cascade of boosted classifiers is executed on the blobs ROIs to detect a LP. The experimental results on our test video with the resolution of 720x576 show that the LPD rate of the proposed system is 99.14% and the average computational time is approximately 42ms.

show abstract

A Fast and Efficient Haar-Like Feature Selection Algorithm for Object Detection

Abstract: 주저자：서강대학교 전자공학과 CAD&ES 연구실,

Cited by 9 publications

References 16 publications

Pengenalan Wajah Menggunakan Implementasi T-shape Mask pada Two Dimentional Linear Discriminant Analysis dan Support Vector Machine

Pengenalan Wajah Menggunakan Implementasi T-shape Mask pada Two Dimentional Linear Discriminant Analysis dan Support Vector Machine

A Study of Face Detection Using CNN and Cascade Based on Symmetry-LGP & Uniform-LGP and the Skin Color

Real-Time Vehicle License Plate Detection Based on Background Subtraction and Cascade of Boosted Classifiers

Contact Info

Product

Resources

About