Computer Vision News - November 2019

2 Summary We Tried for You 8 The first manipulation we perform is thresholding. This manipulation removes undesired marks from the image and sharpens the region of interest. There are many thresholding methods in OpenCV, we shall use the default one. Thresholding is usually working better on grayscale images so we will also convert to black and white image. The code and results look like: "In some cases, it is unavoidable to train our own model." There are several tricks we can do to improve the performance of the Tesseract OCR. This time we will take a car plate that is already aligned. Without preprocessing, we get an empty string as a result, meaning that we were not able to recognize the text. We now go through a few manipulations that can turn the image recognizable. This is what we start from: Preprocessing to improve the accuracy of Tesseract