Advances in Engineering: an International Journal (ADEIJ), Vol.2, No.3 15

RECOGNITION OF SKIN CANCER IN DERMOSCOPIC IMAGES USING KNN CLASSIFIER

K. Ramya Thamizharasi¹, J.Ganesh^{2 1}Final M.Tech, Department Of Information Technology, Dr.Sivanthi Aditanar College of Engineering, Tiruchendur, India ²Assistant Professor, Department Of Information Technology, Dr.Sivanthi Aditanar College of Engineering, Tiruchendur, India

https://airccse.com/adeij/papers/2319adeij02.pdf

ABSTRACT

The largest organ of the body is human skin. Melanoma is a fastest growing & deadliest cancer which starts in pigment cells (melanocytes) of the skin that mostly occurs on the exposed parts of the body. Early detection is vital in treating this type of skin cancer but the time and effort required is immense. Dermoscopy is a non invasive skin imaging technique of acquiring a magnified and illuminated image of a region of skin for increased clarity of the spots on the skin The use of machine learning and automation of the process involved in detection will not only save time but will also provide a more accurate diagnosis. The skin images collected from the databases cannot be directly classified by the automation techniques. The reason is twofold: (a) Lack of clarity in the features which is mainly due to the poor contrast of the raw image and (b) Large dimensions of the input image which causes the complexity of the system. Hence, suitable techniques must be adopted prior to the image classification process to overcome these drawbacks. The first drawback can be minimized by adopting suitable pre- processing techniques which can enhance the contrast of the input images. The second drawback is solved by incorporating the feature extraction technique which reduces the dimensions of the input image to high extent. Further, K-NN (K-Nearest Neighbor) classifier is used for classification of the given image into cancerous or non- cancerous.

KEYWORDS

Dermoscopy , Melanocyte , Histogram

Advances in Engineering: an International Journal (ADEIJ), Vol.2, No.3

REGION CLASSIFICATION AND CHANGE DETECTION USING LANSAT-8 IMAGES

S.Nivetha¹ and Dr.R.Jensi^{2 1}M.E, Final Year Dr.Sivanthi Aditanar College of Engineering, Tiruchendur. ²Assistant Professor, Dr.Sivanthi Aditanar College of Engineering, Tiruchendur.

                                                      https://airccse.com/adeij/papers/2319adeij01.pdf

ABSTRACT

The change detection in remote sensing images remains an important and open problem for damage assessment. A new change detection method for LANSAT-8 images based on homogeneous pixel transformation (HPT) is proposed. Homogeneous Pixel Transformation transfers one image from its original feature space (e.g., gray space) to another feature space (e.g., spectral space) in pixel-level to make the pre-event images and post-event images to be represented in a common space or projection space for the convenience of change detection. HPT consists of two operations, i.e., forward transformation and backward transformation. In the forward transformation, each pixel of pre-event image in the first feature space is taken and will estimate its mapping pixel in the second space corresponding to post-event image based on the known unchanged pixels. A multi-value estimation method with the noise tolerance is produced to determine the mapping pixel using K-nearest neighbours technique. Once the mapping pixels of pre-event image are identified, the difference values between the mapping image and the post-event image can be directly generated. Then the similar work is done for backward transformation to combine the post-event image with the first space, and one more difference value for each pixel will be generated. Then, the two difference values are taken and combined to improve the robustness of detection with respect to the noise and heterogeneousness of images. (FRFCM) Fast and Robust Fuzzy C-means clustering algorithm is employed to divide the integrated difference values into two clusters- changed pixels and unchanged pixels. This detection results may contain few noisy regions as small error detections, and a spatial-neighbor based noise filter is developed to reduce the false alarms and missing detections. The experiments for change detection with real images of LANSAT-8 in Tuticorin between 2013-2019 are given to validate the percentage of the changed regions in the proposed method.

KEYWORDS

Change detection, remote sensing, heterogeneous images, mapping image