The aim of this project is to find a way to scan books at a low cost around 100 $ and make it accessible to everyone. The possibility of filming the book while turning its pages by hand instead of taking a picture of every page will be explored. The aim is to reduce the movement made by the user. There are plenty of scanners found on internet and on the market which have mechanical arms or manual page turning. But generally the cost is high and the setup cumbersome. The project includes two main parts: a hardware one and a software one. The first part concerns the actual construction of the scanning setup. The second part is about obtaining a good image of each page from the video. At the end of the project the idea is to have a plan for building a cheap scanner which anyone could make and a software to treat the video taken during the scanning process. The target user is really wide going from students to countries in development institutions, the aim is to create a cheap do-it-yourself setup with a software to treat the data given by the camera. Also, it is interesting to explore dynamic which is not yet exploited at its full potential.
The aim of creating a low-cost scanner of around 100$ seems attainable. The project consists of two principal parts as said above. The scanner set-up which consists of a stable, light, book holder, etc… And the programming part which will deliver a program to treat the continuous images flow. To do so we will need to gain knowledge in image treatment, but it seems realistic given our knowledge in programming. The other big problem is the construction choices to guarantee the lowest price.
Concerning the latter, the objective is to create a light but stable support, foldable if possible in order to carry it easily at the lowest price. The support includes a lighting system, a fixation for the camera and eventually a device blowing air perpendicularly to flatten the pages once turned. A first reflexion has been made about the materials. We think that cardboard might be the best compromise because of its cost, flexibility and easiness to handle. A second reflexion was made about the shape of the book support. In order to prevent book damages but also to facilitate and accelerate page turning, we consider the V-shape as being the best option. The angle has not yet been precisely defined. The different fixations could be made with wood, plastic or cardboard. The idea is to create our scanner using mostly cardboard, a cheap and original concept.. The size would be as compact as possible, for a book format still to be defined. For the light, we consider installing two powerful and economic LED lights, slightly orientable, on each side of the camera, avoiding imageview field obstruction as well as shadows. If we validate the air blowing option, two spray nozzles would be situated above the light to prevent shadows, perpendicularly to the middle of the page with an orientable exit.
The idea is to provide a plan mentioning the materials to use and indicating how to cut, fold and assemble the scanner. Some elements are still to be decided. Indeed, the whole structure might be in one single piece or separate pieces, for instance one part to support the book and another part supporting all technical devices. In case of commercialisation, the carryable concept is a double advantage, so that the packaging is already optimal.
The main problem of the software part is to deliver a file containing the image of the book’s pages. Since we have decided to scan with a continuous flow of images, we will have to go from video flow to a set of pictures in a pdf. In the best case it would be to make a program doing all the tasks discussed afterwards. Instead of making each part, it would be easier if our program used different open source softwares. Otherwise we could provide a guideline with homemade and open source softwares that the user would need in order to do the conversion.
The program will have to choose the images from the video. It could recognize the best picture by studying similarities between images or by detecting the movements when turning pages, both possibilities have to be explored. Also depending on the camera that we will choose (Gopro, Xiaomi action camera or CMOS sensor) and the number of cameras (1 or 2), we will have to cut the images and to resize them. After those steps, the images might be still a bit rough. So we will try to implement or use a software to assemble multiple images to have at the end the best result possible according to the video.
Since the project is split into two main parts, our teamwork will follow this repartition as well. The first group will work on the physical part of the project and the second group will work on the image treatment part.
Choosing the right design and size (weeks 0-3)
- Implementing the first prototype(s) (weeks 3-7)
- Draw the final plans(weeks 7-12)
- Scanning of books to test the software (weeks 12-13)
- Choice of camera and discussion with the other group (weeks 0-1)
- Searching for existing programmes that could be useful and getting useful knowledge to make the needed program (weeks 1-5)
- Implementing the program to choose and cut the images (weeks 5-10)
- Resizing the images and optionally assembly of multiple images to ameliorate the quality (weeks 10-12)
- Testing the software with the scanner (week 13)
Report and last details (week 14)