projects

Projects

Back to Home

Project 1: Chip Design for a LED Driver and Photoreceiver (Final Project of EE591 Analog Circuit at University at Buffalo)  [Proposal] [Report]

Abstract: In this project we need to design a chip for LED Driver and Photoreceiver. In the LED Driver stage, We apply a 5 stage current starved VCO and achieve a oscillation frequency varies from 830Hz to 213.5kHz that is wider than the require frequency range. A multistage Common Source Amplifier is applied to obtain a high voltage gain, and a push-and-pull power amplifier is used as the output stage. In the photoreceiver block, we use a Type 3 differential amplifier as the High Gain Amplifier which has a CMRR of 86dB and a constant gain up to 200kHz. A two-stage operational amplifier is applied as the threshold detector with a output of 5V or 0V. We skillfully utilize the junction capacitors and bias circuits and thus save a lot of space without laying any capacitors in the chip.  The chip layout is laid by Magic7 software and circuit simulation is done on Spices3 at UNIX server.

Fig 1. Layout of VCO and LED Amplifier for LED Driver Block

 Fig 2. Layout of High Gain Amplifier and Threshold Detection for Photoreceiver Block

Fig 5. Mpeg4 Encoder Software

Project 3: Study of Incomplete Image Inpainting Technology Using Tensor Analysis (Master Project) [PDF] [PPT]

Abstract: In the virtue of our improved Tensor Voting method, I proposed image inpainting algorithms for repairing of small-area and large-area damage respectively. In the small-area repairing algorithm, we built a priority based on the tensor features and then used an iterative method to repair the image window with greatest priority. As the size of the window and thresholds were automatically selected, this algorithm was simple and effective. In the large-area repairing algorithm, our method included both structural and textural repairing processes. That was, first, to repair the missed structure like edges of the entire image, and then repair the missing texture within. Because of the consideration for the missed structure, our algorithm achieved a better visual effect. In the end, I made both subjective and objective assessments to the superiority of the proposed methods over traditional PDE and Criminisi’s exemplar based methods.

 Fig 6. Image Inpainting Result for Soccer

Fig 7. Image Inpainting Result for Bungee

Project 4: Study of 2D EMD in Super-resolution Reconstruction [PDF] [PPT]

Abstract: I finished matlab code for 2D EMD technique and the code is still widely used my lab. Based on that, I proposed a new image zooming method using 2D EMD technique. The main procedures was designed as the following: first, by 2D EMD method, an image is decomposed into several layers, where each of them contains different details or possesses different roughness; second, each layer is zoomed using a combination method with different parameters that are proportional to roughness of that layer; at last, the final zooming result is reconstructed from the combined interpolated layers. Experimental results demonstrate the effectiveness of the proposed method.  

Fig 8. 2D EMD result for cameraman: (a)~(d) are IMF1~IMF4, (e) is the residue term, (f)~(j) are the corresponding residue for the 1st to 5-th decomposition layer respectively.

Fig 11.  Wireless Sensor and Receiver

Project 6: Voltage Controlled LC Oscillator (The 6th National Undergraduate Electronics Design Contest) [PDF] [html]