| 摘要: | 在工業 5.0 的這個時代，雷射加工已是耳熟能詳的加工方式，大多數加工機具都能透過安裝激光頭使用雷射加工，依機台架構不同，常見的有龍門式平台、機械手臂甚至是專為雷射加工設計的反射式平台；本論文使用 H-Bot 機構作為研究平台，基於架構，相較於上述的龍門平台、機械手臂，先天上具有平均馬達負載的優勢；而與反射式平台相比，成本也較為低廉。
而進行研究方便考量，本研究採用 C#人機介面，在加工之前對二維影像進行前處理，將其轉換為機台加工代碼 GCode；藉由 STM32 微控制器解讀加工碼，透過 I/O、RS485 控制馬達運行步數、速度，同時以 PWM 控制雷射輸出功率進行加工。
實驗方面，針對不同雷射功率與燒灼產生灰階之間的關聯，找到其關係曲線，討論其特性，並反饋應用於功率上下限調整、功率分配方式；於影像前處理部分，採用直方圖均化前處理方法提升成品對比度、抖動前處理強化燒灼灰階表現；加工路徑增加了斜向路徑，比對其餘舊有路徑效果，提升加工精度的同時盡可能維持相同加工時間；最後與前人學者羅的加工結果相比，各項指標如峰值訊噪比、結構相似度都顯示具有較好的對比度，且原圖相似度較高。;In the era of Industry 5.0, laser processing is a well-known processing method. Most processing tools can use
laser processing by installing a laser head. Depending on the structure of the machine, common ones include
gantry platforms, robotic arms, and even reflective platforms designed for laser processing. This paper uses the H Bot mechanism as the research platform. Based on the structure, it has the advantage of average motor load
compared to the aforementioned gantry platform and robotic arm. Compared with the reflective platform, the cost
is also lower.
For the convenience of research, this study uses a C# human-machine interface, preprocesses the two dimensional image before processing, and converts it into machine processing code GCode. The STM32
microcontroller interprets the processing code, controls the motor running steps and speed through I/O, RS485,
and controls the laser output power for processing with PWM.
In terms of experiments, the relationship between different laser powers and the grayscale generated by
burning is found, the relationship curve is found, its characteristics are discussed, and feedback is applied to the
adjustment of power upper and lower limits and power distribution methods. In the image preprocessing part, the
histogram equalization preprocessing method is used to improve the contrast of the finished product and the jitter
preprocessing to enhance the performance of the burning grayscale. The processing path has added a diagonal
path, compared with the rest of the old path effects, to improve the processing accuracy while maintaining the
same processing time as much as possible. Finally, compared with the processing results of previous scholars Luo,
all indicators such as peak signal-to-noise ratio and structural similarity show better contrast, and the similarity to
the original image is higher. |
