Remote Sensing Image Classification: A Comprehensive Review and Applications

Table 8

Texture feature-based models.

Title	Methods	Datasets	Accuracy	Limitations
Alexey et al. [145]	Gabor	Corel, Li, Caltech101	83%, 88%, 70%	Increase in computational cost
Song et al. [117]	Wavelet	Corel 1K, Corel 5K, Corel 10k	99%, 56%, 35%	Increase in computational cost
Zhu et al. [118]	Histogram	Corel 1K, Corel 5K	87%	–
Phadikar et al. [146]	MPEG-7 edge detector	NUSWIND	98%	Computational cost increased by using GA
Sharmila and Sharmila [147]	Canny edge detector	Corel10K	68%	Running cost increased due to large image input
Cao et al. [121]	DWT,EDH	Corel	73.50%	No ML used in experimentation
Deselaers et al. [122]	Ranklet transformation	Corel 5K, Corel 10K	67.4%, 67.9%	Computational cost increased due to multiple dimensions of features
Zhang et al. [123]	GLCM	Corel 5K	66.90%	No optimization algorithm is used to reduce computational complexity
Ren et al. [124]	Gabor filter	Corel 5K	79%	Computational cost increased because of feature dimensions
Mu et al. [125]	DWT	Corel	90%	High computational time