Comparative evaluation of change detection techniques based on multispectral images for measuring land cover dynamics of mango (Mangifera indica L.) in Malihabad, Lucknow (UP)

  • Harish Chandra Verma ICAR-Central Institute for Subtropical Horticulture, Lucknow-226101, India.
  • Tasneem . Ahmed Advanced Computing &Research Laboratory, Department of Computer Application, Integral University, Lucknow-226026, India.
  • Shailendra . Rajan ICAR-Central Institute for Subtropical Horticulture, Lucknow-226101, India.
Keywords: Accuracies, change detection, Landsat 8 OLI, mango (Mangifera Indica L.), satellite images, vegetation


Change detection is a process of identifying and quantifying the differences between images of the same scene at different times and in mango fruit crop is still a very challenging task. The major challenge of change detection in mango is discriminating between perennial fruit crops because many of these crops have similar reflectance profiles. Accurate change detection in the mango area will help the government prepare for area expansion and conservation planning. The main aim of this study was to determine the efficient change detection method for mango fruit crop among the most commonly used change detection methods. In this work, a comparative study was conducted by using Landsat 8 OLI images of two different dates i.e. 14 February, 2015 and 25 February, 2019 of the Malihabad mango region of Lucknow district. In this paper, four change detection methods namely, Vegetation Index Differencing (VID), Log Ratio (LR), Principal Component Analysis (PCA), and Image Rationg (IR) were evaluated to detect the changes in mango crop area. To extract the mango regions, Soil Adjusted Vegetation Index (SAVI) images of year 2015 and 2019 were calculated and further used to retrieve the VID, LR, PCA, and IR raster images. After that, these four raster images were thresholded to annotate the ‘Positive change’, ‘Negative change’ and ‘No change’ areas; thereafter, to obtain the final change map, masking was applied to mask out the non-mango area. Change detection accuracy was calculated using ground truth data to assess performance. After conducting the comparative analysis of all four change detection methods, it was found that the highest change detection accuracy is achieved with the VID and PCA followed by LR and IR, respectively. PCA and VID methods provided higher accuracies, followed by LR to detect changes in mango crop area. It is due to these methods' capability to enhance the information on the change.