Topic > Alignment using binary and grayscale images - 2166

1.6.3 Alignment using binary and grayscale images Method 1 This first algorithm uses information from binary and grayscale images to estimate the angle of inclination. It is based on the binary image filtering algorithm 1.2.1, the Sobel edge detection filter and the classical Hough transform. Since we are looking for angles between -25 and 25 degrees, the window length is set to 3 and the threshold to 2 for the filtering algorithm. If a white pixel meets the conditions of the filtering algorithm, we then apply the Sobel filter edgedetection at the considered point of the grayscale image. If the gradient amplitude is greater than 255, votes are performed in all directions in the ac -cumulator. Peaks in the accumulator are detected using the method proposed in 1.5 .2.Method 2This method differs from the previous one only in the voting scheme. In fact, instead of voting in all directions, the gradient directory is used to calculate the estimated inclination angle at the point considered using (1.12). To maintain accuracy, votes are performed between θ − 2◦ and θ + 2 ◦ .To increase the accuracy of the algorithm for a poorly cropped image, it might be interesting to vote in more directions than -2 or 2 degrees. However, this implies greater computation time and perhaps not greater accuracy. 1.7 Results For the experiment, 25 documents from magazines, business letters and annual reports were considered. Documents are tested using predefined angles between 0 and 25 degrees. The following table provides the mean (M), standard deviation (SD) and computation time (T) in seconds for the different proposed methods. These tests were performed on a Pentium 4......center of the paper......they were done on 12 randomly chosen words or groups of words. Finally, boldness could be estimated by looking at the estimated variation in boldness between words on the same line. It is also mandatory to note that the estimated boldness varies depending on the characters of the words.2.4 Scaling Algorithms2.4.1 Scale2xScale2x is a real-time graphics effect that can increase the size of small bitmaps by guessing missing pixels without interpolating pixels and blurring the images. was originally developed for the AdvanceMAME project in 2001 to improve the quality of older games with low video resolution. Also available are Scale3x and Scale4x derivative effects that scale the image by 3x and 4x (8). Image upsampling is calculated by applying some rules to each pixel of the input image. First let's consider the following matrix 3 × 3: