Optical character recognition. I did this by using tesseract software in Linux. The script was simple enough to apply OCR on all images.
ls *.png>pics; let count=228 while read line; do tesseract $line $count.odt -l eng; ((count++)); done<pics
tesseract is such a nice tool, 95% of text was accurate. For more detail, read this https://mandeep7.wordpress.com/2015/10/16/scan-documents-via-mobile-camera-ocr/