Region merging using color histogram similarity(Bhattacharya coefficient) on a mean-shift segmented image
조회 수: 1 (최근 30일)
이전 댓글 표시
Hello!
For my project, I need to get an initial image segmentation using mean-shift algorithm, followed by two iterations of region merging using color histogram similarity (Bhattacharya coefficient). I have been able to get the segmented image using EDISON system but I am not sure how to proceed with the next steps as i am new to matlab. I am not sure how to label the segments in the segmented image.
I have included the steps mentioned in the study I am following. It is given below:
"In human visual perception, several regions with similar color or spatial adjacency should be merged into one region. These regions need to be represented by several feature descriptors and a rule for region merging must be defined. We employ the method of color histogram similarity to calculate the similarity between two adjacent regions.Each color histogram is quantized into 16 levels and then into a total of 4096 bins in each region. The color histogram similarity ρis calculated between two adjacent regions (e.g. regions Pand Q) using Bhattacharyya coefficient. We use the region adjacency graph to store the similarity of the pair of regions.
After calculating the similarity between two neighbor regions, we use the Region Adjacency Graph (RAG) to store the similarity of the pair of regions. In this work, the merging rule is defined as follows:
if the similarity (Bhattacharyya coefficient) between adjacent regions P and Q is the highest among all the similarities, then regions Pand Q are merged.
Following this merging rule, we obtain the final segmentation map."
I also need to find the first largest and the second largest segments in the final map. Also, I am unable to understand the meaning of the part I have marked as bold.
If anyone can guide me through, I shall be very grateful.
댓글 수: 0
답변 (0개)
참고 항목
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!