Issn 2091-5446 ilmiy axborotnoma научный вестник scientific journal
The Proposed UzSL Recognition System
Download 1.19 Mb. Pdf ko'rish
|
ilovepdf merged
- Bu sahifa navigatsiya:
- Fig.2.
The Proposed UzSL Recognition System
As shown in Fig (2), all training and testing images are collected and features are then extracted by applying SIFT algorithm which is described above. The output of the SIFT algorithm is feature vectors with high dimension for each image. Fig.2. Block diagram of the proposed system ILMIY AXBOROTNOMA INFORMATIKA 2021 - yil, 1 - son 110 Processing these high dimension vectors takes more CPU time. So, we need to reduce the dimensions of the features by applying dimensionality reduction method, such as LDA which is one of the most suitable methods to reduce the dimensions and increase the separation between different classes. LDA also decreases the distance between the objects belong to the same class. The feature extraction and reduction are performed in both training and testing phases. The matching or the classification of the new images are only done in the testing phase. The proposed system, as illustrated in Fig (2), is mainly consists of two phase: Training and Testing. – Training Phase: The detailed steps of training phase of proposed algorithm can be summarized as follows. 1. Collecting all training images (i.e. gestures of Uzbek Sign Language). 2. Extracting features using SIFT from each images. 3. Representing each image by one feature vector. To reduce the numberfeatures; we used LDA as a dimensionality reduction method. – Testing Phase: The detailed steps of testing phase of proposed algorithm can be summarized as follows. 1. Collecting the testing image, 2. Extract the features of testing images and convert them to featurevectors. 3. Feature vectors are projected on LDA space to reduce their dimensionality. 4. Matching or classifying the test feature vector to identify final decision (i.e. identify the character corresponding to this image (sign)). Download 1.19 Mb. Do'stlaringiz bilan baham: |
Ma'lumotlar bazasi mualliflik huquqi bilan himoyalangan ©fayllar.org 2024
ma'muriyatiga murojaat qiling
ma'muriyatiga murojaat qiling