Method and system for extended depth of field calculation for microscopic images

Patent metadata

Field	Value
Publication number	US-9897792-B2
Application number	US-201314437174-A
Country	US
Kind code	B2
Filing date	Nov 4, 2013
Priority date	Nov 30, 2012
Publication date	Feb 20, 2018
Grant date	Feb 20, 2018

Abstract

The invention relates to an image processing method and system for constructing composite image with extended depth of field. The composite image may be constructed from a plurality of source images of a scene stored in an image stack. The method includes aligning the images in the image stack such that every image in the image stack is aligned with other images in the stack, performing illumination and color correction on the aligned images in the image stack, generating an energy matrix for each pixel of each illumination and color corrected image in the image stack by computing energy content for each pixel, generating a raw index map that contains the location of every pixels having maximum energy level among all the images in the image stack, generating degree of defocus ma and constructing the composite image.

First claim

We claim:

1. An image processing method for constructing a composite image with extended depth of field, from a plurality of source images of a scene stored in at least one image stack, the plurality of source images being taken at substantially identical fields of view, the method comprising:
aligning the images in the image stack such that every image in the image stack is aligned with other images in the stack, wherein the images are aligned by a bi-directional image alignment method, the bi-directional image alignment method includes,
arranging the images in a sequence in the image stack;
selecting a first reference image, such that the first reference image is the substantially central image in the image sequence;
aligning the first reference image with immediate left side and immediate right side images in the sequence; and
repeating the alignment process for the next image on left side and right side of the reference image with respect to the aligned images;

performing illumination and color correction on the aligned images in the image stack;
generating an energy matrix for each pixel of each illumination and color corrected image in the image stack by computing energy content for each pixel;
generating a raw index map that contains the location of every pixels having maximum energy level among all the images in the image stack;
generating degree of defocus map by comparing the energy content at a particular pixel in all the images against a reference signal and repeating the process for all the pixels; and
constructing the composited image using the raw index map and the degree of defocus map.

2. The method as claimed in claim 1 , wherein the left side and right side images are aligned by a hierarchical coarse-to-fine parametric image alignment process.

3. The method as claimed in claim 2 , wherein the hierarchical coarse-to-fine parametric image alignment process is implemented by a guide optimization process.

4. The method as claimed in claim 1 , wherein the step of performing illumination and color correction comprises:
selecting a second reference image and a sample image from the aligned images such that the second reference image and the sample image are consecutive images in the image stack;
converting the selected images from RGB colour space to HSV colour space;
computing an average luminance and an average saturation value of the HSV colour space images;
calculating the percentage deviation of the average luminance and the average saturation for the first sample image with respect to the second reference image;
multiplying the first sample image with a luminance correction factor, if the percentage deviation of the average luminance is more than a predefined threshold level, and/or multiplying the first sample image with a saturation correction factor, if the percentage deviation of the average saturation variation is more than a predefined threshold level to obtain the modified first sample image, else using the first sample image as modified first sample image;
repeating the process for all the images, by considering the modified first sample image image as the first reference image and next consecutive image as first sample image; and
converting the modified images from HSV color scale to RGB color scale.

5. The method as claimed in claim 4 , wherein the predefined threshold level is more than 2 percent deviation of the average luminance and/or the average saturation variation.

6. The method as claimed in claim 4 , wherein the luminance correction factor is the ratio of the average value of the illumination of the first sample image divided by average value of illumination for the second reference image.

7. The method as claimed in claim 4 , wherein the saturation correction factor is the ratio of the average value of the saturation of the first sample image divided by average value of saturation for the second reference image.

8. The method as claimed in claim 1 , wherein the step of generating an energy matrix comprises:
converting each color and illumination corrected RGB image to grayscale image;
down sampling the image to low resolution of intensity values in the range of 0 to 1; and
convolving the down sampled image with a complex wavelet filter bank to generate the energy matrix.

9. The method as claimed in claim 1 , wherein the reference signal is a Gaussian curve is generated by:
taking an array of energy values from energy each matrices in temporal direction for each pixel location and identifying a peak value in each array;
identifying a local minima on both sides of the Peak energy, value, in the array; and
generating the Gaussian curve by taking log of the energy values between the maxima and two minima.

Assignees

• L&T Technology Services Ltd

Inventors

• Rout Rabi
• Desai Vajendra
• Ramaraju Harikrishnan
• Sukumaran Gineesh
• Ganesh Sistu
• Priyadarshi Devvrata
• Abraham Eldho
• Mishra Saurabh
• Kumar Sahoo Sudipta
• Tripathi Nivedita
• Goutham Siddhartha
• Gauraw Kunal

Classifications

• G06T2207/20221

  Image fusion; Image merging · CPC title

• G02B21/367 Primary

  providing an output produced by processing a plurality of individual source images, e.g. image tiling, montage, composite images, depth sectioning, image comparison · CPC title

• H04N23/689

  Motion occurring during a rolling shutter mode · CPC title

• G06T2207/10052

  Images from lightfield camera · CPC title

• H04N1/40012

  Conversion of colour to monochrome · CPC title