DIGITAL HISTOPATHOLOGY AND MICRODISSECTION
A computer implemented method of generating at least one shape of a region of interest in a digital image is provided. The method includes obtaining, by an image processing engine, access to a digital tissue image of a biological sample; tiling, by the image processing engine, the digital tissue image into a collection of image patches; identifying, by the image processing engine, a set of target tissue patches from the collection of image patches as a function of pixel content within the collection of image patches; assigning, by the image processing engine, each target tissue patch of the set of target tissue patches an initial class probability score indicating a probability that the target tissue patch falls within a class of interest, the initial class probability score generated by a trained classifier executed on each target tissue patch; generating, by the image processing engine, a first set of tissue region seed patches by identifying target tissue patches having initial class probability scores that satisfy a first seed region criteria, the first set of tissue region seed patches comprising a subset of the set of target tissue patches; generating, by the image processing engine, a second set of tissue region seed patches by identifying target tissue patches having initial class probability scores that satisfy a second seed region criteria, the second set of tissue region seed patches comprising a subset of the set of target tissue patches; calculating, by the image processing engine, a region of interest score for each patch in the second set of tissue region seed patches as a function of initial class probability scores of neighboring patches of the second set of tissue region seed patches and a distance to patches within the first set of issue region seed patches; and generating, by the image processing engine, one or more region of interest shapes by grouping neighboring patches based on their region of interest scores.
1 - 45 . (canceled)
46 . A digital image processing method for separating foreground objects from a background scene, the method comprising:
obtaining, via at least one processor, a digital image of a scene;
tiling, via the at least one processor, the digital image of the scene into a set of image patches;
assigning, via the at least one processor, each image patch of the set of image patches an initial class probability score generated by a trained foreground object classifier for each image patch;
generating, via the at least one processor, a first set of patches from the set of image patches and having initial class probability scores satisfying a first criteria;
generating, via the at least one processor, a second set of patches from the set of image patches and having initial class probability scores satisfying a second criteria;
calculating, via the at least one processor, a region-of-interest (ROI) score for each patch in the second set of patches as a function of initial class probability scores of neighboring patches of the second set of patches and a distance to patches within the first set of patches; and
generating, by the at least one processor, one or more ROI shapes representing foreground objects by grouping neighboring patches based on their ROI scores.
47 . The method of claim 46 , wherein the digital image of the scene comprises video game image data.
48 . The method of claim 46 , wherein the digital image of the scene comprises augmented reality image data.
49 . The method of claim 46 , wherein the digital image of the scene comprises image data from a vehicle.
50 . The method of claim 49 , wherein the image data from the vehicle comprises image data from an autonomous vehicle.
51 . The method of claim 50 , further comprising avoiding a collision with at least one of the foreground objects based on the one or more ROI shapes.
52 . The method of claim 46 , wherein the digital image comprises a frame of a video.
53 . The method of claim 46 , wherein the trained foreground object classifier comprises a trained neural network.
54 . The method of claim 46 , wherein the one or more ROI shapes comprise foreground masks.
55 . The method of claim 54 , wherein the foreground masks separate a foreground from a background in the digital images.
56 . The method of claim 46 , wherein the set of image patches comprises patches of uniform size and shape.
57 . The method of claim 56 , wherein the set of image patches comprises patches of at least 256 pixels by 256 pixels.
58 . The method of claim 57 , wherein the set of image patches comprises patches of at least 400 pixels by 400 pixels.
59 . The method of claim 58 , Wherein the set of image patches comprises patches of at least 1000 pixels by 1000 pixels.
60 . The method of claim 46 , wherein the set of image patches comprises patches of non-uniform size or shape.
61 . The method of claim 46 , wherein the trained foreground object classifier includes at least one of the following classifiers: a support vector machine, softmax, decision tree, random forest, k-nearest neighbor, linear and quadratic discriminant analysis, ridge regression, multilayer perceptron (MLP), Hyper-pipes, Bayes net, k-means clustering, or naive Bayes classifier.
62 . The method of claim 46 , further comprising obtaining access to a database of a priori trained foreground object classifiers.
63 . The method of claim 62 , further comprising selecting the trained foreground object classifier according to classifier selection criteria based on metadata bound to the digital image.
64 . The method of claim 46 , wherein tiling the digital image includes filtering the set of image patches based on color channels of pixels within each the image patch of the set of image patches.
65 . The method of claim 64 , further comprising filtering the set of image patches as a function of variance with respect to the color channels.
66 . A system for separating foreground objects from a background scene, the system comprising:
at least one non-transitory computer-readable memory storing software instructions; and
at least one processor coupled with the at least one memory and that performs the following operations upon execution of the software instructions:
obtaining a digital image of a scene;
tiling the digital image of the scene into a set of image patches;
assigning each image patch of the set of image patches an initial class probability score generated by a trained foreground object classifier for each image patch;
generating a first set of patches from the set of image patches and having initial class probability scores satisfying a first criteria;
generating a second set of patches from the set of image patches and having initial class probability scores satisfying a second criteria;
calculating a region-of-interest (ROI) score for each patch in the second set of patches as a function of initial class probability scores of neighboring patches of the second set of patches and a distance to patches within the first set of patches; and
generating one or more ROI shapes representing foreground objects by grouping neighboring patches based on their ROI scores.
67 . A non-transitory computer-readable medium storing software instructions thereon for separating foreground objects from a background scene which, when executed, cause at least one processor to perform operations comprising:
obtaining a digital image of a scene;
tiling the digital image of the scene into a set of image patches;
assigning each image patch of the set of image patches an initial class probability score generated by a trained foreground object classifier for each image patch;
generating a first set of patches from the set of image patches and having initial class probability scores satisfying a first criteria;
generating a second set of patches from the set of image patches and having initial class probability scores satisfying a second criteria;
calculating a region-of-interest (ROI) score for each patch in the second set of patches as a function of initial class probability scores of neighboring patches of the second set of patches and a distance to patches within the first set of patches; and
generating one or more ROI shapes representing foreground objects by grouping neighboring patches based on their ROI scores.