Using DSP to implement portable low vision aid
| Autor: | Jin-Wen Wu, 吳健聞 |
|---|---|
| Rok vydání: | 2008 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 96 Objective：A special designed vision-assisted system for low vision subjects has characters with technique of image enhancement so that it can improve the subjects’ visual acuity without much magnification. Meanwhile it can also provide user an interface to adjust the magnification of the image when reading paper or book. Methods：The causes of low vision in humans are quite different, thus visual-function evaluation is required for each low vision subject prior to put on this customer-made vision-assisted device. The protocol of the evaluation is described as follows: Contrast sensitivity function (CSF) is carried out in subjects and the results are compared with the average normal CSF which is obtained from tens of normal subjects without any history of ocular diseases. Visual degradation transfer function (VDTF) is then calculated for each subject according to the normal CSF so that the visual deficiency in spatial frequency of the low vision subject can be defined. To verify the flow of processing image enhancement, we used Matlab to analyze the still image and estimated video performance by Simulink. If this algorithm of simulation was perfect, then we could embed the algorithm into DSP to implement a portable low vision aid. The implementation of the low-vision aid was carried out through TI’s (Texas Instrument) evaluation board of TMS320DM642 (720MHz) that outputs to a pair of head-mounted display. Results：We have built several models of algorithm in Matlab and Simulink, by which image qualities are compared through the examination of these models. Two of these models (constant multiplier and variable multiplier methods) were selected and the results of calculations were verified without error occurrence (after testing, the way of variable multiplier seemed to be better than the constant multiplier). After optimization for increasing the program efficiency, requirement of real time image display can be achieved while runs on the DSP device finally. It provides a complete interface for the users to do adjustment for obtaining clear image and suitable magnification. Conclusion：This study not only provides a method of image processing in software to enhance the contrast of captured image with characteristics for particular low-vision subjects, it also results in an implementation of a real hardware of low-vision aid with sufficient functionality and quality. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|