Content extraction from advertisement display boards utilizing Region growing algorithm

In recent years portable camera devices have gained increased popularity and embedded visual processing, text extraction from natural scene images like advertisement display boards, government office boards has become a key problem in everyday lives. The problem is challenging in nature due to variations in the font size and color, text alignment, illumination change and reflections. Here a novel method for extraction of text from advertisement display boards using Region growing algorithm is proposed. The proposed algorithm is generally composed of four stages, the colored image is converted into grayscale image and canny edge method is used to detect the edges of the image, the edge detected image is preprocessed by applying morphological operations and rule based method is used to remove the non text objects based on width, height and area, later finding the centroid point of the connected component of identified objects and finally proposed algorithm region growing method is used to start extracting the characters. The method is robust and insensitive to noise, blur, variation in font size and style, color, uneven thickness, and varying lightning conditions. The text extraction accuracy of 90.94% is achieved.