This week is the third week of the year 2 project. This week we have made a breakthrough and we have worked out the most part of of projecct.
In the third week, the hardware part is mostly done and we have placed the both camera on the board with a fixed distance which provides the convenience for further alignment by software. Besides the hardware progress, our software part has more significant results!
Our software team has been continuously working on the picture fusion too. The first progress is that we have successfully convert the decimal temperature infomation into a RGB picture. This means we can now have the intuitive display on our Raspberry Pi screen. The picture below shows how it was converted to real picture.
Display from Raspberry Pi Decimal temperature given by thermal camera
In addition, we worked out the way to improve the software to allow camera have a higher refresh rate at this time. At the beginning our thermal can only take the picture every 3 seconds which is far from what we expected as "Real Time Fusion". Through a large amount of experimentation and modification of our code, we found it's our improper code causes thermal camera not output the information in real time. After debug our output rate is raised to 3.5Hz which we are satisfied now. Beyond that, we also worked out an GUI version of the software to allow the user friendly using experience.
However, we are now expericing bottleneck too, our current photo fusion technology is computationally intensive. It runs very slowly even on our laptop, and uses a lot of memory as well, it usually taking around 10 seconds to process a photo which pull down the whole performance. Obviously our Raspberry Pi is far less powerful than our computer, which means it takes longer to process the same photo. Therefore, we need continue to find a possible solution to make it process faster.
GUI version of our software(wating for fusion function)
Overall, this week's task completion is out of our expected, we haven't got an working solution to our project. In the lask week, we must working even harder to find the solution to support image fusion on Raspberry Pi and prepare for our presentation.
Summary of week’s activities:
Placed the camera on board with fixed distance
Convert the decimal thermal information to visible picture.
Improve thermal camera refresh rate
GUI version software build
Problem found:
1. The current image fusion algorithm takes too long for Raspberry Pi to process.
2. Image fusion is not working as we want, the edge alignment is still not working.
Tasks for next week:
1. Combine the thermal image and visible image. 2. Improve the code to allow faster processing. 3. Integrate all functions to the GUI version and run-on Raspberry Pi.
Progress check:
✔ Experiment Material Preparation
✔ Raspberry Pi Built
!!! Combine thermal and visible picture together (OUT OF DATE)
✔ Improve the picture resolution
⃞ Poster making
Comments