lz 这个应当有多种思路
因为只要求最大8个值及对应坐标 所以用排序 过了点
设图像前8个值所组成的1-D数组为MAX_value，对应1-D坐标数组为MAX_coor_x，MAX_coor_y,threshold为MAX_value的中各元素的最小值，A为对应图像，i,j为宽，高,k为栈指针,threshold_coor为threshold在MAX_value中对应的位置(防止有多个最小值，所以可能是个数组)，threshold_coor_multi为有多个最小值的情况下，相同值的个数,0表示各不相同。
Pseudo:
i,j initialization
for i,j scan every pixel of the image except the first 8 pixels
%%in-loop initialization
gray value with coordinates of the first 8 pixels to MAX_value
calculate threshold_coor_multi in MAX_value
coordinates of the first 8 pixels to MAX_coor_x, MAX_coor_y
threshold_coor = the order no. of smallest in MAX_value
threshold = smallest in MAX_value
%%in-loop computation
if A[i,j] >=threshold
then
A[i,j] to MAX_value[threshold_coor]
i,j to MAX_coor_x[threshold_coor],MAX_coor_y[threshold_coor]
if A[i,j] = threshold %%case without variation of thresold
then
threshold_coor_multi++
update threshold_coor
break
end if
if threshold_coor_multi >0 %%case for equal smallest gray values
then
threshold_coor_multi --
update threshold_coor
else
update threshold
update threshold_coor
end if
end for
个人观点 仅供参考
[本帖最后由 silas_xue 于 2009-3-27 07:00 编辑]