초록 ▼

A method of segmenting regions of an image wherein a number of partitions are determined based on a range of an image histogram in a logarithmic luminance domain. Regions are defined by the partitions. A mean value of each region is calculated by K-means clustering wherein the clustering is initiali

A method of segmenting regions of an image wherein a number of partitions are determined based on a range of an image histogram in a logarithmic luminance domain. Regions are defined by the partitions. A mean value of each region is calculated by K-means clustering wherein the clustering is initialized, data is assigned and centroids are updated. Anchor points are determined based on the centroids and a weight of each pixel is computed based on the anchor points.

대표청구항 ▼

1. A method of segmenting regions of an image comprising the steps of: determining a number of partitions based on a range of a histogram of the image;defining regions by the partitions;estimating a centroid of each region;assigning pixel data to the estimated centroids by assigning each pixel data

1. A method of segmenting regions of an image comprising the steps of: determining a number of partitions based on a range of a histogram of the image;defining regions by the partitions;estimating a centroid of each region;assigning pixel data to the estimated centroids by assigning each pixel data point to its nearest centroid to form a plurality of clusters;estimating new centroids by taking the mean of each cluster; and,determining anchor points based on the new centroids. 2. The method of claim 1 wherein the centroid is determined by K-means clustering applied in a logarithmic luminance domain. 3. The method of claim 2 wherein anchor points are based on a half distance between adjacent centroids. 4. The method of claim 3 further comprising the step of computing a weight of each pixel within each region. 5. The method of claim 4 further comprising tone mapping the image based on the pixel weighting. 6. A method of segmenting regions of an image comprising the steps of: determining a number of partitions based on a range of an image histogram in a logarithmic luminance domain;defining regions by the partitions;calculating a mean value of each region by K-means clustering wherein the clustering is initialized, data is assigned and centroids are updated;determining anchor points based on the centroids; and,computing a weight of each pixel based on the anchor points. 7. The method of claim 6 wherein the cluster is initialized by one of a random seed method or a random partition method. 8. The method of claim 7 wherein the centroids are updated by an iterative process. 9. The method of claim 6 wherein regions are permitted to overlap. 10. The method of claim 6 further comprising the step of normalizing the weights. 11. The method of claim 6 further comprising the step of binarizing the weights. 12. A method of segmenting regions of an image comprising the steps of: designing an energy function based upon a probability of anchor points in an image histogram;iteratively differentiating the energy function using a chain rule to determine the anchor points; and,computing a weight of each pixel in the image based on the anchor points. 13. The method of claim 12 wherein the energy function is expressed as: J=∑i⁢pLi⁡[(1-ⅇN1)×ⅇN1ⅇN1+ⅇN2+⁢…⁢+ⅇNk+(1-ⅇN2)×ⅇN2ⅇN1+ⅇN2+⁢…⁢+ⅇNk+⁢…⁢+  ⁢(1-ⅇNk)×ⅇNkⅇN1+ⅇN2+⁢…⁢+ⅇNk]where ⅇNj=ⅇ-[S⁡(L1Aj)-0.5]2σ2⁢ (j=1, 2, . . . , k) and where i is an index of bins in the histogram, Li is a corresponding luminance value, PL is a probability of Li, {Aj} are the anchor points, σ is a constant, and S(•) is a saturation function, defined by S⁡(x)={1x>1x1/γotherwise. 14. The method of claim 13 wherein eNj is a Gaussian curve which defines a measurement for exposure. 15. The method of claim 14 further comprising a gradient descent method applied at the differentiating step.