Test result not that accurate as you'd showed?

[447983454]

sorry to bother you, I'm trying to run your code while the disparity result I get is not that accurate like you'd shown. take "cone" picture(LLL, RRR.jpg) as an example, their edge is not distinct in my case. but some of your results are clear and accurate. Does it concern arguments setting or etc? looking for your reply! sincerely!

[xxsong5]

hi, there is one main point which can explain you different results: 1. such as cone, which has rich texture, my result with distinct edge was gotten by using Absolute-Difference matching cost and not by the Absolute-Difference-Structure matching cost which is published in this code. because ADS has better performance than AD in lower-texture situation, through a lot of test. So matching cost is the key for different situation. by the way, result with distinct edge of cone is the result (with AD matching cost) before interpolation and non-peak[ Peak(disparity_l, 2, 0); Interpolation(disparity_l);]

[447983454]

thanks a lot! I really learn a lot from your reply!