- 學位論文
基於人類視覺生理特徵的LED照明控制研究
The study of LED lighting control based on physiological characteristic of human vision
摘要
照明研究在節省能源上是一個重要課題。目前固態照明已經成為引導照明系統的顯學。自白光LED發明以來，已經成為一種對節能有重大影響的研究。白光LED的驅動方法有許多種，本論文專注於pulse-width modulation (PWM)操作LED的設計當作主題，針對給視覺使用的照明做研究。本論文所探討的視覺，除了人類視覺之外還探討模仿人類設計而來的機器視覺。 人類的視覺很難度量。本論文整理了過去百年來眼科與視覺醫學方面進行過的研究，挑選適合給照明設計利用的理論。根據視覺生理學理論，可以整理成三態即閃爍、閃爍融合、穩態。同樣的，機器視覺也有類似的三態構造，但機器視覺直接受到應用要求的不同而有所不同，較容易理解。根據整理完成的相關理論，供給之後的實驗做應用。 為了能夠尋找適用於普遍照明系統的設計，不採用醫學級精密測試實驗，而使用可控的PWM產生器與LED lamp，安裝於半封閉燈箱以人眼觀察進行實驗。實驗環境加入兩種環境光源進行，目的是模擬室內照明。進行三種實驗：閃爍實驗尋找最高刺激度的參數、閃爍融合實驗尋找使閃爍感消失的區域、穩態照明實驗測試人眼不能分辨的最小duty差。本實驗可得一個總結，製作警告燈時最佳頻率範圍在7Hz至15Hz間；照明燈為200Hz以上，duty調整間隔為10%為最佳。 在早期的實驗發現，LED在PWM驅動下呈現的亮度並非完全線性，而一些論文從電流對亮度的反應一樣能夠支持這樣的論點。非線性亮度的問題，在機械視覺，如自動化生產線檢查就造成困難。因此，設計一個自動化data logger來進行線性度驗證。採用最簡單的LED驅動電路避免任何可能的影響，將LED lamp與照度計封進一個暗箱，用PC下指令操作PWM的頻率與duty後等待穩定，讀取照度並留下紀錄。依此流程完成50 KHz 到 1 KHz，duty 間隔為10%的實驗。實驗完成後發現最線性的區域落在4KHz 到3KHz 之間。由於照明設備與選用的LED不會一樣，但可以利用此方法進行線性度校正。 最後，本論文完成了對於人類視覺的最佳照明驅動，以及對機器視覺的最佳照明驅動方法。在未來如果能將兩者合一探討，將會是另一個有趣的議題。
關鍵字
none並列摘要
Lighting research on saving energy is an important subject. The solid-state lighting has become a famous doctrine to guide the lighting system. The White-light LED has a significant effect on energy saving, since the invention has become an important research. There are many kinds of White-light LED driver method, this dissertation focus on Pulse-Width Modulation (PWM) operation LED lighting, based on human vision characteristics. In this dissertation, not only discuss human vision, also discuss machine vision that mimics human vision. Human vision is very difficult to measure. In this dissertation, we analyzed researches conducted over the past hundred years the medical aspects of ophthalmology and visual, selected the suitable theory for lighting system design. According to the theory of visual physiology, it may be organized into three architectures which are flashing, flicker fusion, and the steady state lit. Similarly, machine vision has a similar structure and three architectures. Machine vision and human vision have some differences. The machine vision character is directly by requirements of the application. According to the related consolidation theory, could supply the experiment and application. And design PC software to expand the data logger function. In order to find suitable for general lighting system design, we do not use the medical level precision test experiment. We use a controllable PWM generator with LED lamp, mounted into semi-enclosed box, with the human eye to observe the phenomenon and then finish the experiments. Experiment environment use two light sources, the purpose is to simulate the indoor lighting. Three experiments: flashing experiment find stimulating the highest degree of parameters; flicker fusion experiments for flashing feeling disappear area; and steady state lighting test to the human eye can't distinguish the minimum differential duty of the PWM. The summary of experiment: the best warning light frequency located between 7 Hz to 15 Hz; lights for more than 200 Hz, duty adjustment between 10% are the best. Found in early experiment, the relation of LED brightness to PWM driven is not completely linear, and some papers propose the driven current response to brightness can support this argument. Nonlinear brightness in machine vision, such as automatic product checking, will be a hard problem. Therefore, we use the data logger and PC software control, verifies the LED linearity. In order to avoid any circuit effect, we use the simplest driver circuit. LED lamp and brightness meter closed in a small dark room, use PC command PWM generator to set frequency and duty, after waiting for stability, read the intensity of illumination and record. In this experiment process, automatic adjust frequency from 50 KHz to 1 KHz, duty form 90% to 10%. The experiment results showed, the linear area is between 4 KHz to 3 KHz. Due to the lighting equipment and LED are different with our experiment, but can use the method of linearity correction. Finally, this dissertation completed the best lighting for human vision drive, and the best lighting for machine vision driven approach. In the future, if one can explore both, it will be another interesting topic.