Muhammad Atif

Ph.D. Thesis: Title: A small structured light 3D camera of high speed and high accuracy for 3D Modeling Abstract: Structured Light 3D camera is gaining attention due to its potential for the high-speed accurate depth measurement. Stereo systems are replaced in the industry by the structured light 3D camera system due to deprived performance in the poor ambient conditions and fail to recover 3D information from the texture-less surface. The challenge is to obtain the depth information accurately for the moving object in the poor ambient conditions. The focus of this dissertation research is to develop a system that can obtain accurate depth information with high scan speed and estimate the camera exposure settings automatically in real time. In this research high accuracy of the depth information is achieved by reducing the outliers by precisely determining the occluded regions and accurate decoding of the patterns, high scan speed is achieved by hardware approach, projecting the patterns with high speed by using Field Programmable Gate Array (FPGA) with commercial video projector and embedded projector. Scan speed 3Hz is achieved and scan time is decreased from 2300msec to 320msec by synchronizing the projector and camera. The camera exposure settings are estimated through the pixel intensity slope which makes the 3D camera system independent from the manual control in the industry environment. The exposure settings are estimated by using a white and ambient frame which makes the system real time because white and ambient are mandatory for 3D reconstruction.

ME Thesis: Title: Characterization of DAC Non-Linearities inside ΣΔ ADC Loop Abstract: My studies into this project lead me to the conclusion that Nyquist-rate converters exhibit two major significant limitations: the anti-alias filter and the quantization noise. At a glance, the requirements of an anti-alias filter are fairly straightforward: the pass-band must accurately pass the desired input signals and stop-band must attenuate any interferer outside the pass-band. The second limitation of Nyquist-rate converters is the quantization noise. Oversampling converters are introduced to overcome the limitation of Nyquist rate data converters. Oversampling data converters offer several advantages over traditional converters. ΣΔ ADC’s are quite famous among oversampled ADC’s due to its characteristics of noise shaping and high resolution. ΣΔ ADC’s loop constitutes quantizer, summer, integrator and digital to analog converter (DAC). These components are non ideal/linear in nature but fortunately non linearities introduced by the quantizer; summer and integrator are corrected by the ΣΔ loop. DAC non linearities cannot be corrected by the ΣΔ loop which causes decrease in the SNR, SNDR and SFDR of the signal. For an n-bit ΣΔ ADC at OSR of 256 a 4-bit flash ADC is required along with n-bit linear DAC. The n-bit DAC linearity cannot be achieved due to CMOS scaling and component mismatching. The main purpose of this thesis is to model the non linearities of ΣΔ loop components and to observe there effects on signal’s SNR, SNDR and SFDR. The methodology involved modeling of ADC’s and DAC’s non-ideal behavior individually and then in tandem. Simulation results show that the non idealities introduced by the ADC have been corrected by the ΣΔ loop but the DAC degrades the signals dynamic parameters (SNR, SNDR, and SFDR).

Final year project: JPEG Image compression algorithm implementation