|
|
 |
| |
|
|
|
A Statistical Model Based on Deep Learning and the Chi-Square Algorithm for Segmenting Grayscale Images |
|
|
|
PP: 1225-1234 |
|
|
doi:10.18576/jsap/150504
|
|
|
|
Author(s) |
|
|
|
Najla M. Aloraini,
|
|
|
|
Abstract |
|
|
| Recently, the necessity to arrive the information precisely extracted from images has demanded, and the want for features and properties extraction from a raw image using segmentation faster and more effectively. There are naturally many techniques to perform segmentation extending from image properties. Deep learning currently provides effective solutions to numerous issues in many applied fields. In this paper, we present a CHi-squared Automatic Interaction Detector (CHAID) in a decision tree algorithm to estimate the precision of various key threshold nodes of an image. Segmentation and merging procedures are auxiliary tools for building a decision tree (DT) and obtaining the appropriate target for segmentation. The statistical learning model is used to predict the appropriate segmentation values based on the dependent variable representing the pixel values of the input image. To measure the accuracy of the model we apply the various common accuracy measurements, such as MASE, MAPE, and sMAPE on tested images. Furthermore, the details of the paper, such as the procedures used, the data sets, and the results achieved are comprehensively evaluated and identified. |
|
|
|
|
|
|
|
