View All Documents & Correspondence
An Approach For Change Detection Method To Remote Sensing Images Using Deep Cnn
Abstract: Change detection (CD) is an important technique for analysing land surface changes using Earth observation data and for identifying connections between social and natural events in geoscience. The vital problem in remote sensing images is to improving the quality of the binary CD map. Binary CD seeks to identify changes and no changing regions; in this work, a supervised deep learning (DL)-based change detection technique was developed to produce an accurate change map. In this invention, we proposed deep convolutional neural network (DCNN) for the detection of identical change between two SAR (Synthetic-aperture radar) images. A convolutional neural network (CNN) is used to extract RSI image characteristics and observed changes into the proper categories. The NASA satellite image dataset has been used for investigation. The total accuracy is around 95%, and networks should learn to differentiate between manufactured and genuine changes, given that these particular artificial alterations are tagged as database updates. As a result, the implementation of the change detection technique is supervised and may be used with any pre-trained CNN or Classification algorithm for classification tasks.
Notices, Deadlines & Correspondence
Patent Information
Applicants
1. VIT-AP UNIVERSITY
Inventors
1. Chafle.Pratiksha Vasantrao
2. Dr. Neha Gupta
Claims
1. A hybrid feature extraction and selection approach for classification which can detect the change between two satellite image objects using proposed CNN.
2. The system mentioned in the claim 1, wherein the image processing is used for filtration, resizing, and normalization to elimination of misclassified instances and the various hybrid feature extraction techniques can able to eliminate overfitting problems and improve the overall accuracy.
3. The system mentioned in the claim 1, wherein the proposed hybrid Deep CNN classification algorithm could efficiently work on validation of various time zone images.
4. The proposed methodology mentioned in the claim 1, wherein provides better and effective results than various state-of-art methods which is used for change detection of SAR images.
Specification
1/We Claim:
1. A hybrid feature extraction and selection approach for classification which can detect the change between two satellite image objects using proposed CNN.
2. The system mentioned in the claim 1, wherein the image processing is used for filtration, resizing, and normalization to elimination of misclassified instances and the various hybrid feature extraction techniques can able to eliminate overfitting problems and improve the overall accuracy.
3. The system mentioned in the claim 1, wherein the proposed hybrid Deep CNN classification algorithm could efficiently work on validation of various time zone images.
4. The proposed methodology mentioned in the claim 1, wherein provides better and effective results than various state-of-art methods which is used for change detection of SAR images.
Documents
Application Documents
| # | Name | Date |
|---|---|---|
| 1 | 202241000120-Form9_Early Publication_03-01-2022.pdf | 2022-01-03 |
| 2 | 202241000120-Form5_As Filed_03-01-2022.pdf | 2022-01-03 |
| 3 | 202241000120-Form3_As Filed_03-01-2022.pdf | 2022-01-03 |
| 4 | 202241000120-Form28_Educational Institution_03-01-2022.pdf | 2022-01-03 |
| 5 | 202241000120-Form2 Title Page_Complete_03-01-2022.pdf | 2022-01-03 |
| 6 | 202241000120-Form1_As Filed_03-01-2022.pdf | 2022-01-03 |
| 7 | 202241000120-Form18_Examination Request_03-01-2022.pdf | 2022-01-03 |
| 8 | 202241000120-Drawings_As Filed_03-01-2022.pdf | 2022-01-03 |
| 9 | 202241000120-Description Complete_As Filed_03-01-2022.pdf | 2022-01-03 |
| 10 | 202241000120-Correspondence_Eligibility Document_03-01-2022.pdf | 2022-01-03 |
| 11 | 202241000120-Correspondence_As Filed_03-01-2022.pdf | 2022-01-03 |
| 12 | 202241000120-Claims_As Filed_03-01-2022.pdf | 2022-01-03 |
| 13 | 202241000120-Abstract_As Filed_03-01-2022.pdf | 2022-01-03 |
| 14 | 202241000120-FER.pdf | 2022-05-13 |
Search Strategy
| 1 | 202241000120E_12-05-2022.pdf |