An unsupervised hair segmentation and counting system in microscopy images
Download as DOCX, PDF0 likes175 views
This paper presents an unsupervised system for hair segmentation and counting in microscopy images, addressing key challenges such as the removal of bright spots, overlapping hairs, and the limitations of conventional detection algorithms. A hair-bundling algorithm is introduced to improve accuracy in counting concealed hairs. The proposed hair counting method outperforms Hough-based techniques, demonstrating robustness to various image distortions.
An unsupervised hair segmentation and counting system in microscopy images
- 1. AN UNSUPERVISED HAIR SEGMENTATION AND COUNTING SYSTEM IN MICROSCOPY IMAGES ABSTRACT This paper focuses on the development of medicalsoftware for clinical applications using advanced image processingalgorithms. Three critical issues of hair segmentation andcounting are addressed in this paper. First, the removal of anybright spots due to oil or moisture, which generate circularpatterns in the middle of the hair and significantly affect theaccuracy of determining the line. Second, two contacting oroverlapping hairs are recognized and counted as a single hair.To solve this problem, we proposed a hair-bundling algorithm tocalculate any concealed hairs. Finally, hairs may be wavy or curly,making the conventional Hough-based line detection algorithmunsuitable, since it suffers from parameter selections, such asthe minimum length of line segment, and distance between linesegments. Our proposed hair counting algorithm is substantiallymore accurate than the Hough-based one, and robust to curls,oily scalp, noise-corruption, and overlapping hairs, under variouswhite balance.