features with explanation

[Mina Naghshnejad]

at first the background is set to black by using the given mask file,
the first column is the # of components: extracted by using graythresh,medfilt2, bwlabel
column 2 to 7 : statmoments: mean, standard diviation, skewness, kurtosis, uniformity=sum(imhist.^2/num of histogram bins), entropy
column 8 to 10: 3 first column of the result of ( fourierfx(fourierf(fourierf(fourierf(gray scale image)))) fourierfx: rowwise fft, fourierf: columnwise fft, real part
column 11 to 13: 3 first column of the result of ( fourierfx(fourierf(fourierf(fourierf(gray scale image)))) fourierfx: rowwise fft, fourierf: columnwise fft, imaginary part
column 14 to 33: radonf is radon transform of image with angle 0,45,90,135 convolved in 5bin histogram. radonf(fourierf(fourierf(fourierf(image)))).
column 34 is the target vector.

this set of feature reached 72.77 with taking 0.7 of data as training with knnclassify , k=8.

Regards,
Mina Naghshnejad