Quantcast
Channel: Machine Learning
Viewing all articles
Browse latest Browse all 62845

It seems like the first step toward good machine vision is scale/distance logic

$
0
0

Is this the current favored technique?

Background: Natural or man made (Color, texture, pattern)

Judging Distance to camera: Visible texture, Atmospheric diffusion

Judging distance to background: Shading, lighting

Texture: Metallic - shiny, machine scraped texture, silvery or coppery. Particulates? Fibrous?

Big or small: Diffusion, visible texture pattern, lighting/shading, blurred close-up effect, using logic relative to size of background

Man Made: Perfect/straight lines + Metallic. Perfect curved surface

It seems like judging scale and distance is the very first step to determining what it is you're looking at. And it seems the best way to do that is to analyze diffusion, texture, camera effects (Blurring), and reasoning out the distance of the background and the object.

Is this currently done?

submitted by xebo
[link][comment]

Viewing all articles
Browse latest Browse all 62845

Trending Articles