Performance Analysis of Color K-Means and Range Filter for Text Detection in Images or Video

Abstract

Text extraction in images and videos is an imperative role in image study and computer vision. The text identification process determines the presence of textual contents in the given input. The main challenges that occurred during the text identification process are complex background, illumination, low-resolution, different color, and variation in font size. In this paper, the color k-means is employed to raise the gap among textual  and non-textual information. A range filter is employed to identify possible text components. Aperture concept has established to identify the actual text sections. The accuracy of the provided model was estimated using the common datasets MRRC, hua's, and nus dataset.. The outcome shows that the approach is promising and encouraging.