To find area of interest from image

Hi all. I have a problem for getting the interested roi from image. They are three dimensional image. I want to remove unnecessary parts from that and make them new one. But the interested area on image sometimes vertical sometimes not. there are two examples there. how can i determine this. After finding this im going to apply lbp there. Or apply to lbp to all image after gray level convertion. Then calculate the pixel values and get rid of the background. The original image and wanted part for extarcting lbp values is attached thanks for suggestions

[Image Analyst]

" there are two examples there. " <=== Where???

You've attached a single 2-D color image. Is your 3-D image a color image or a gray scale volumetric image? How can you apply a local binary pattern to a 3D image? For a single pixel there are 26 neighbors in 3D so you'd have a possible 2^26 number of patterns. Tell me why it even makes sense for you to do this. What do you want to measure or learn from your image?

[ali]

Sorry. I just posted one image and border with lines on image that is the interested region for me. This is rgb image first I convert to gray level. Then I want to find the differentiation lbp values from that picture especially vertical place that is bordered.

[ali]

I only extract features with LBP from the border of vertical line or closest regions like in that figure. Because background is the same for all other figures.

Answer 1

[ali]

Actually I want to seperate background and object very well before apply LBPto images. What kind of threshold should effective on this.

Answer 2

[ali]

The image is like that I want only vertical line with inside points other background should remove. How can do this effectively.