RGB Image Cluster Evaluation for Human Blood Group Identification by MLP Classification
Shridevi Soma1, Pooja Yashwantaray2

1Dr. Shridevi Soma, Professor, Department of Computer Science and Engineering, Pooja Doddappa Appa College of Engineering, Kalaburagi (Karnataka), India.
2Pooja Yashwantaray, M.Tech., Department of Computer Science and Engineering, Poojya Doddappa Appa College of Engineering, Kalaburagi, (Karnataka), India.

Manuscript received on 12 August 2019. | Revised Manuscript received on 18 August 2019. | Manuscript published on 30 September 2019. | PP: 4197-4202 | Volume-8 Issue-3 September 2019 | Retrieval Number: C6005098319/2019©BEIESP | DOI: 10.35940/ijrte.C6005.098319
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Image processing is helping researchers to reach their goals in many ways, especially in medical fields. Blood organization is very important when it comes to receiving a blood exchange. The most important blood group identification method is ABO blood group system and the RhD blood group system. Blood groups are defined by the occupancy or preoccupied of a specific agglutinate on the get around of a red blood cell. Identifying the blood group is very important for medical treatment in pathological tests, at some point it gives us an inaccurate and also expensive result, therefore, to overcome these problems an efficient and optimal solution is required. The need for accurate detection is high in a disaster situation where there are no laboratory people or experts available to detect the type of it. In the proposed method, we have collected 50 blood sample images for each of 8 blood groups, total 400 blood sample images are considered for experimentation. In preprocessing, the median filter is used to eliminate noise from the blood images. Then these images are converted from RGB to grayscale conversion and also resizing of the images is carried out. Region based segmentation by using two methods Markov Random Field and Region Adjacency Graph are used for segmentation, texture, color, and shape features are extracted from segmented images. Hence this paper proposes a pixel cluster based analysis of the blood type based on the pixel analysis features. The overall accuracy of blood group determination is 93.85%.
Keywords: ABO Blood Image, Preprocessing, Region Based Segmentation, Feature Extraction, MLP Classifier.

Scope of the Article: Classification