In one embodiment, a method includes obtaining an image comprising a plurality of pixels, determining, for a particular pixel of the plurality of pixels, a gradient value, classifying, based on the gradient value, the particular pixel into a flat class or one of a plurality of edge classes, and denoising the particular pixel based on the classification.

An imaging device 10 according to one aspect of the present invention includes: a subject distance acquisition section 115; a movement amount acquisition section 120 that acquires an amount of movement of the subject on the basis of the subject distance; a restoration processing determination section 125 that determines, on the basis of the amount of movement acquired by the movement amount acquisition section 120, whether the restoration processing should be performed on the images through a restoration filter, a restoration strength of the restoration processing should be adjusted and the restoration processing should be performed on the images, or the restoration processing should not be performed on the images; and a restoration processing execution section 105 that performs the restoration processing on the images through the restoration filter or with the adjusted restoration strength, on the basis of the determination of the restoration processing determination section 125.

An imaging device 10 according to one aspect of the present invention includes: a subject distance acquisition section 115; a movement amount acquisition section 120 that acquires an amount of movement of the subject on the basis of the subject distance; a restoration processing determination section 125 that determines, on the basis of the amount of movement acquired by the movement amount acquisition section 120, whether the restoration processing should be performed on the images through a restoration filter, a restoration strength of the restoration processing should be adjusted and the restoration processing should be performed on the images, or the restoration processing should not be performed on the images; and a restoration processing execution section 105 that performs the restoration processing on the images through the restoration filter or with the adjusted restoration strength, on the basis of the determination of the restoration processing determination section 125.

In one embodiment, a computer-based method includes obtaining a first image where the first image includes one or more patterns, generating a second image that substantially removes or reduces the one or more patterns from the first image at least partially by automatically detecting the one or more patterns and a zone where the one or more patterns occur in the first image, converting the first image to frequency domain data, applying a multi-parameter filter to the frequency domain data to substantially remove or reduce the one or more patterns. The parameters may include bandwidths in a depth and azimuthal direction. The parameters may be adapted in the multi-parameter filter based on the one or more patterns. The method also includes transforming the frequency domain data to spatial domain data and outputting the second image based at least in part on the spatial domain data.

In one embodiment, a computer-based method includes obtaining a first image where the first image includes one or more patterns, generating a second image that substantially removes or reduces the one or more patterns from the first image at least partially by automatically detecting the one or more patterns and a zone where the one or more patterns occur in the first image, converting the first image to frequency domain data, applying a multi-parameter filter to the frequency domain data to substantially remove or reduce the one or more patterns. The parameters may include bandwidths in a depth and azimuthal direction. The parameters may be adapted in the multi-parameter filter based on the one or more patterns. The method also includes transforming the frequency domain data to spatial domain data and outputting the second image based at least in part on the spatial domain data.

A method is described that includes executing a convolutional neural network layer on an image processor having an array of execution lanes and a two-dimensional shift register. The executing of the convolutional neural network includes loading a plane of image data of a three-dimensional block of image data into the two-dimensional shift register. The executing of the convolutional neural network also includes performing a two-dimensional convolution of the plane of image data with an array of coefficient values by sequentially: concurrently multiplying within the execution lanes respective pixel and coefficient values to produce an array of partial products; concurrently summing within the execution lanes the partial products with respective accumulations of partial products being kept within the two dimensional register for different stencils within the image data; and, effecting alignment of values for the two-dimensional convolution within the execution lanes by shifting content within the two-dimensional shift register array.

A method is described that includes executing a convolutional neural network layer on an image processor having an array of execution lanes and a two-dimensional shift register. The executing of the convolutional neural network includes loading a plane of image data of a three-dimensional block of image data into the two-dimensional shift register. The executing of the convolutional neural network also includes performing a two-dimensional convolution of the plane of image data with an array of coefficient values by sequentially: concurrently multiplying within the execution lanes respective pixel and coefficient values to produce an array of partial products; concurrently summing within the execution lanes the partial products with respective accumulations of partial products being kept within the two dimensional register for different stencils within the image data; and, effecting alignment of values for the two-dimensional convolution within the execution lanes by shifting content within the two-dimensional shift register array.

A computing device may obtain an input image. The input image may have a white point represented by chrominance values that define white color in the input image. Possibly based on colors of the input image, the computing device may generate a two-dimensional chrominance histogram of the input image. The computing device may convolve the two-dimensional chrominance histogram with a filter to create a two-dimensional heat map. Entries in the two-dimensional heat map may represent respective estimates of how close respective tints corresponding to the respective entries are to the white point of the input image. The computing device may select an entry in the two-dimensional heat map that represents a particular value that is within a threshold of a maximum value in the heat map, and based on the selected entry, tint the input image to form an output image.

A computing device may obtain an input image. The input image may have a white point represented by chrominance values that define white color in the input image. Possibly based on colors of the input image, the computing device may generate a two-dimensional chrominance histogram of the input image. The computing device may convolve the two-dimensional chrominance histogram with a filter to create a two-dimensional heat map. Entries in the two-dimensional heat map may represent respective estimates of how close respective tints corresponding to the respective entries are to the white point of the input image. The computing device may select an entry in the two-dimensional heat map that represents a particular value that is within a threshold of a maximum value in the heat map, and based on the selected entry, tint the input image to form an output image.

An image processing part includes an edge enhancing part, an artifact detecting part and a compensating part. The edge enhancing part emphasizes an edge portion of an object in input image data. The artifact detecting part detects a corner outlier artifact at an area adjacent to the edge portion of the object. The compensating part compensates the corner outlier artifact. Accordingly, the edge portion of the object may be enhanced and the corner outlier artifact is decreased so that the display quality may be improved.