# US 2025/0095159 A1 · Automated region selection for auto sweep > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2025/0095159 A1 - Document ID: US20250095159A1 - Status: Published application - Kind: A1 - Application number: 18/470,348 - Filed: 2023-09-19 - Published: 2025-03-20 - Earliest claimed: 2023-09-19 - Assignees: FEI Company - Applicants: FEI Company - Inventors: Umesh ADIGA; Derek HIGGINS; Mark BIEDRZYCKI; Melanie DOUD - Web page: https://patentgrep.com/patent/US20250095159A1 ## CPC - G06T 7/136: involving thresholding - G06T 7/11: Region-based segmentation - H01J 37/153: Electron-optical or ion-optical arrangements for the correction of image defects, e.g. stigmators - H01J 37/21: Means for adjusting the focus - H01J 37/222: {Image processing arrangements associated with the tube} - H01J 37/265: {Controlling the tube; circuit arrangements adapted to a particular application not otherwise provided, e.g. bright-field-dark-field illumination} - G06T 2207/10061: from scanning electron microscope - G06T 2207/20016: Hierarchical, coarse-to-fine, multiscale or multiresolution image processing; Pyramid transform - G06T 2207/30148: Semiconductor; IC; Wafer - H01J 2237/1532: Astigmatism - H01J 2237/216: Automatic focusing methods - H01J 2237/221: Image processing - H01J 2237/28: Scanning microscopes - H01J 2237/31749: Focused ion beam - G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL - H01J: ELECTRIC DISCHARGE TUBES OR DISCHARGE LAMPS (spark-gaps H01T; electric arc lamps H05B31/00; particle accelerators H05H) ## Abstract Sample regions of interest (ROIs) for use in autofocus procedures are identified based on a gradient image of the sample. ROIs with gradient values greater that a threshold are selected, and eigenvalues of the associated image matrices are determined. ROIs with suitable variation in eigenvalues such as at least two relatively large eigenvalues are associated with high contrast features orientated in multiple directions so that such ROIs are suitable for automatic focus and astigmatism correction. Suitable ROIs can also be identified based on a histogram of gradient orientations. ## Claims ### Claim 1 A method, comprising: obtaining a sample image; processing the sample image to obtain gradient magnitudes; and selecting a set of initial candidate regions of interest (ROIs) in the sample image associated with gradient magnitudes greater than a threshold value. ### Claim 2 (depends on 1) The method of claim 1, further comprising ranking each of the initial candidate ROIs based on the associated gradient magnitudes. ### Claim 3 (depends on 1) The method of claim 1, further comprising evaluating the initial candidate ROIs based on a histogram of gradient orientations (HGO) associated with each of the initial candidate ROIs. ### Claim 4 (depends on 3) The method of claim 3, wherein the gradient orientations in the HGO are based on gradient orientations determined from nearest neighbor pixels associated with each pixel in the candidate ROIs. ### Claim 5 (depends on 3) The method of claim 3, wherein the evaluating the initial candidate ROIs based on the histogram of gradient orientations (HGO) is performed by comparing the HGO of each ROI to a uniform HGO. ### Claim 6 (depends on 1) The method of claim 1, further comprising: processing the sample image to produce images associated with at least three image resolutions corresponding to an image pyramid; and ranking the initial candidate ROIs based on eigenvalues associated with corresponding image blocks in at least one of the three images of the image pyramid. ### Claim 7 (depends on 6) The method of claim 6, further comprising selecting a first set of candidate ROIs from the set of initial candidate ROIs based on the rankings of the candidate ROIs of the initial set of candidate ROIs obtained with the lowest resolution image of the image pyramid. ### Claim 8 (depends on 7) The method of claim 7, further comprising selecting a second set of candidate ROIs based on rankings of the first set of candidate ROIs set obtained with an image of the image pyramid having a resolution other than the lowest resolution. ### Claim 9 (depends on 1) The method of claim 1, further comprising: processing the sample image to produce a Gaussian image pyramid that includes images associated with at least three image resolutions; and for each image of the Gaussian image pyramid, beginning with the lowest resolution image and ending with the highest resolution image, ranking candidate ROIs from the set of initial candidate ROIs based on eigenvalues associated with corresponding image blocks obtained from the respective image of the Gaussian image pyramid, wherein rankings based on a lower resolution image of the Gaussian image pyramid are used to select candidate ROIs for processing by a next higher resolution image of the Gaussian image pyramid. ### Claim 10 (depends on 9) The method of claim 9, wherein the rankings are based on eigenvalue magnitudes. ### Claim 11 (depends on 9) The method of claim 9, wherein the rankings are based on magnitudes of N eigenvalues having largest magnitudes, wherein N is an integer that is less than 36 and greater than 2. ### Claim 12 (depends on 9) The method of claim 9, wherein the rankings are based on at least one of: a ratio of a sum of magnitudes of N eigenvalues having largest magnitudes to a sum of magnitudes of all eigenvalues, a ratio of a sum of magnitudes of two eigenvalues having the largest magnitudes to the sum of magnitudes of all eigenvalues, a ratio of magnitudes of the eigenvalues having the largest magnitudes, and ratios of each of the N eigenvalues having the largest eigenvalue to an average of the N eigenvalues having the largest eigenvalues. ### Claim 13 (depends on 12) The method of claim 12, wherein the rankings are based the ratio of the sum of the magnitudes of the N eigenvalues having largest magnitudes to the sum of magnitudes of all eigenvalues. ### Claim 14 (depends on 11) The method of claim 11, wherein the images associated with at least three image resolutions have resolutions corresponding to M/2K, wherein M is a size of the sample image and K is an integer greater than zero and less than 10. ### Claim 15 (depends on 11) The method of claim 11, further comprising selecting a final set of candidate ROIs based on eigenvalues associated with image blocks in the highest resolution image, wherein each candidate ROI of the final set of candidate ROIs is selected so that: a ratio of a sum of the magnitudes of the N eigenvalues having largest magnitudes to either a sum of magnitudes of all eigenvalues or a sum of magnitudes of all eigenvalues except the N eigenvalues having the largest magnitudes, wherein the ratio is greater than 0.7; and a ratio of magnitudes of the eigenvalues having two largest magnitudes is between 0.5 and 2. ### Claim 16 (depends on 15) The method of claim 15, wherein a block size for image blocks selected for each image of the Gaussian image pyramid is different. ### Claim 17 (depends on 15) The method of claim 15, wherein the Gaussian image pyramid is obtained with a Gaussian kernel having a standard deviation corresponding to between 3 and 10 image pixels. ### Claim 18 (depends on 15) The method of claim 15, further comprising adjusting a charged-particle-beam (CPB) optical column based on at least one image associated with a ROI in the final set of candidate ROIs. ### Claim 19 (depends on 18) The method of claim 18, further comprising making a series of adjustments of the CPB optical column based on a plurality of different ROIs from the final set of candidate ROIs. ### Claim 20 (depends on 19) The method of claim 19, wherein the plurality of different ROIs is selected beginning with a highest ranking and continuing in a sequence of declining rankings. ### Claim 21 A charged-particle beam (CPB) imaging system, comprising: a CPB optical column operable to direct a CPB to a sample and produce an image of at least a portion of a sample; and a system controller coupled to the CPB optical column and operable to shape and direct the CPB to the sample, wherein the system controller is configured to: process a sample image to obtain a gradient magnitude image; select a set of initial candidate regions of interest (ROIs) in the sample image associated with gradient magnitude image portions having gradient magnitudes greater than a threshold value; ranking each of the initial candidate ROIs; and selecting a ROI from the initial candidate ROIs based on the ranking, and adjusting focus and astigmatism based on the selected ROI. ### Claim 22 (depends on 21) The CPB imaging system of claim 21, wherein the system controller is configured to evaluate candidate ROIs based on images from a set of images defined as an image pyramid. ### Claim 23 (depends on 22) The CPB imaging system of claim 22, wherein the system controller is configured to evaluate candidate ROIs based on eigenvalues associated with the candidate ROIs in the images of the image pyramid. ### Claim 24 (depends on 22) The CPB imaging system of claim 22, wherein the rankings are determined based on ratios of magnitudes of eigenvalues associated with the candidate ROIs in the images of the image pyramid. ### Claim 25 (depends on 21) The CPB imaging system of claim 21, wherein the system controller is configured to evaluate each candidate ROI based on an associated histogram of gradient orientations (HGO). ### Claim 26 (depends on 25) The CPB imaging system of claim 25, wherein the gradient orientations in the HGOs are based on gradient orientations determined from nearest neighbor pixels associated with each pixel in the candidate ROIs. ### Claim 27 (depends on 25) The CPB imaging system of claim 25, wherein the evaluating the initial candidate ROIs based on the histogram of gradient orientations (HGO) is performed by comparing the HGO of each ROI to a uniform HGO.