8/12/2019 Conference Poster 3

1/1

ADVANCEDDOCUMENTSCANNERWITHDIGITALMOSAIC

Sameer Desai, Steven Ghodke & Sidharth Kulkarni | Dept. of E&TC

samades@gmail.com, steven.ghodke@gmail.com, sksidforyou55@gmail.com

Project Supervisor: Prof. Vinayak B Kulkarni

Objectives

Our main objective is to achieve image scanthrough phototransistor array, instead of theusual CCD methodology. We also aim to include

the feature of creation of a digital mosaic frommultiple scanned pictures, which can have ap-plication or use in scanning very big sizes/areasand will be a Unique Selling Point or the stand-out feature of the project.

Introduction & Mechanism

We built an array of phototransistors and at-tached them to an X-Y plotter consisting of 4rods to scan the entire target area vertically &horizontally. Light reflected by the glass plateon which the input object is kept is detected byphototransistors, which then is given to the Cfor A-D conversion. It is then sent to MATLABby serial communication. MATLAB codes arethen used for the digitization & for mosaicing.Our mechanism is a variation of the CIS tech-nology presently in use.

Figure 1: Scanning mechanism

Parameters Speed : 15 mins for full scan of A2 (approx)

Sensitivity : upto 5mm

Threads of rotation (pitch) : 2mm

Image size & resolution : User controllablethrough MATLAB

Maximum scan size : A2

Conclusion

Main advantages are ease of implementation,Design flexibility, scope for further develop-ments & mosaicing/extra large document scan-ning ability. But at the same time, using pho-totransistors involves a certain drop in perfor-mance with respect to the speed of scanning &output quality. Also, use of black & white scan-ning though ideal for document scanning too re-duces the quality due to the low bit depth &inability to scan detailed colors.

References

[1] William D Clark. Document scanner mechanism.IBM Journal of Research and Development, 23 ,1998.

[2] R Modugno, A Ferrante, E Stasolla, and S Impe-dovo. New trends in digital scanning processes. 10thInternational Conference on Document Analysis and

Recognition, 2009.

Block Diagram

The block diagram of our project is shown in Figure 2 on the left. It shows the flow of data, fromwhen the scanner is first started. Stepper motors which are driven by L297 drivers are used to drivethe X-Y plotter & phototransistors act as the sensor. Then on the right, it also shows the actual

working circuit of our project.

Figure 2: Block Diagram (left); Actual working circuit (right)

ResultsScanning Results-

The scanner presently needs fairly large sized characters in order for the text to be read & identifiedproperly. The output scanned for the signature of one of us is as shown in the Figure 3.

Figure 3: Scanning Output

Black & white scanning is considered the most ideal for projects & documents. This is so, as theresolution of the output image is more user controllable as compared to color scanning due to thelesser bit depth involved. Lesser bit depth also reduces the scanner complexity thus increasing easeof implementation. Mosaicing Results-For testing our MATLAB mosaic code,we took the photograph of one of us & also the signaturescanned previously & split it into 2 halves.We took the 2 halved images as the input for our MATLABprogram. The end points of both the images from where they were to be stitched were selected. Thiswas done as shown in Figure 4. The end point selection is crucial, as improper selection can lead toa distorted or in worst case completely wrong output. So, it is vital to precisely select the stitchingpoints for seamless stitching.

Figure 4: Left half input (left); Right half input (right)

After the end points are selected the 2 halves are stitched back along the designated stithing p ointsto give the mosaiced output. The output mosaiced image & its comparison with the actual image &is as shown below in the Figure 5.Similarly multiple such scans can be mosaiced together to form one final output image, which in ourcase will be the output scan of the large sized document.

Figure 5: Mosaicing Output