Home       Chase County Overview       Makesig       Maxlike       Mindist   Piped       Cluster       Isocluster       Results       References

  


Mindist


Mindist uses a minimum distance to means classification based on information contained in a set of signature files. The mean reflectance on each band for a signature determines the minimum distance to means classification. Pixels are grouped by classes with the closest mean values. Band-space distances are normalized to justify differences in the fluctuating signatures.

Using the maxlike operation, I chose raw as the distance type and indicated an infinite maximum search distance. For the classification, I entered the crops, forest, grassland, pavement, soil, and water signature files previously created. I gave the resulting image an output file name and titled it Mindist Classification.


This is the resulting image of the maxlike classification. Using this classification technique there are a few noticeable errors with the resulting image. I believe many of the pixels classified as bare soil are actually rock/pavement and many of the pixels classified as rock/pavement are actually bare soil. The mindist classification did not do a good job of separating out agricultural crops from other vegetation along the Cottonwood River. Also, it appears that some of the bare soil was misclassifed as water bodies.


This pie graph displays a statistical summary of the maxlike classification results.




  
Home       Chase County Overview       Makesig       Maxlike       Mindist   Piped       Cluster       Isocluster       Results       References