Counting the Amount of Spermatozoa Active Perframe Video Using Morphology and Local Adaptive Threshold

I Gede Susrama (Student Doctoral Program of Electrical Engineering, Institut Teknologi Sepuluh November, Surabaya, Indonesia)
Tobing L (Health Polytechnic of Surabaya, Indonesia)
I Dewa Gede Hari Wisana (Health Polytechnic of Surabaya, Indonesia)
I Ketut Edi Purnama (Associate Proffessor, Dep of Multimedia and Network Eng, Institut Teknologi Sepuluh November, Surabaya, Indonesia)
Mauridy Harry Purnomo (Professor of Electrical Enggineering, Institut Teknologi Sepuluh November Surabaya, Indonesia)

Nusantara Science and Technology Proceedings
International Seminar of Research Month Science and Technology in Publication, Implementation and Co

Sperm analysis is a preliminary examination performed on male infertility cases, one of which is determining normal and abnormal motility, performed by the expert. This sperm analysis can also be done automatically with computer-assisted, that is by taking per-frame sperm video then done segmentation process. In the process of segmentation there are some constraints such as video data taken has different intensity, so it takes some segmentation method. This study compares the process of Morphology and Local Adaptive Threshold (LAT) process, with the same preprosessing process using Otsu segmentation, so that from some segmented sperm video frame data will be known which segmentation should be used. From the result of the research, it is found that the intensity of light in the sperm frame is analyzed evenly, morphology can be used and if the sperm image in uneven intensity analysis will be more effective using Local Adaptive Threshold (LAT), thus to perform better feature extraction, but from some frames of sperm that were randomly tested obtained more accuracy using morphology process that is 82% compared to using local adaptive threshold process that is only 52%.