Gondal, Muhammad Adil2018-08-062018-08-062018https://escholar.umt.edu.pk/handle/123456789/3118Supervised by: Mr. Nabeel Sabir BalochPakistan Sign Language to Natural Language Page 5 of 50 Formatted: Justified, Tab stops: 6", Right Formatted: Font: 10 pt Formatted: Font: 10 pt Formatted: Font: 10 pt Formatted: Font: 10 pt ABSTRACT In this thesis, Pakistan Sign Language to Natural Language System has been proposed. To conquer any hindrance between deaf and an ordinary individual, we need to think of a framework, which can make an interpretation of one dialect to another. The effective communication method between deaf community and many other dictionaries of words have been defined to make the communication possible is Sign Language. A gesture is recorded by Kinect, which gives in-depth vision image and color vision image of everything in front of it. We extract only depth stream from it as we need only skeletal data in our case. By Extracting, feature vector against every gesture and after its normalization and classification the English text word against each gesture will be generated as a result. For this we have proposed a system which takes an PSL gesture as an input and will produce its comparable English word. Our proposed framework uses a few tools and methods, as after normalization we concluded a frame descriptor. When it comes out to be last gesture of frame, we summarize all the saved frame descriptors by summarizing them. Each row represents a frame, and this is called Feature Vector. After doing this process each row will represent a gesture and our classifier to build model, so it can be used for prediction can use this data. Last but not the least a English word is created from this method.enEffective communication methodFrame descriptor, PSL gestureBSThesis