Example method includes: transmitting a plurality of probe images from an IoT device at an edge network to a server hosting a target DNN, wherein the plurality of images are injected with a limited amount of noise to probe sensitivities of the target DNN to the red, green, and blue colors; receiving a feedback comprising a plurality of DCT coefficients unique to target DNN from the server hosting the target DNN; computing a plurality of color conversion weights based on the feedback received from the server; converting a set of real-time images from RGB color space to YUV color space using the plurality of color conversion weights unique to the target DNN; compressing the set of real-time images using a quantization table unique to the target DNN by the IoT device; and transmitting the compressed set of real-time images to the server hosting the target DNN for DNN inferences.

There is provided an information processing apparatus configured to obtain criterion data, which is reference image data, by executing a adjustment parameter determining program, obtain evaluation data which is image data scanned by a scanner, generate a plurality of pieces of comparison data by adjusting the evaluating data using a plurality of adjustment parameters different from each other, respectively, associate each comparison data with an adjustment parameter used to generate the comparison data, and display the criterion data and the plurality of pieces of comparison data in a list. When a selecting operation of selecting one from the plurality of pieces of comparison data is performed, the information processing apparatus determines an adjustment parameter for the scanner based on the adjustment parameter associated with the selected comparison data.

Apparatuses, systems, and methods related to an image processor formed in an array of memory cells are described. An image processor as described herein is configured to reduce complexity and power consumption and/or increase data access bandwidth by performing image processing in the array of memory cells relative to image processing by a host processor external to the memory array. For instance, one apparatus described herein includes sensor circuitry configured to provide an input vector, as a plurality of bits that corresponds to a plurality of color components for an image pixel, and an image processor formed in an array of memory cells. The image processor is coupled to the sensor circuitry to receive the plurality of bits of the input vector. The image processor is configured to perform a color correction operation in the array by performing matrix multiplication on the input vector and a parameter matrix to determine an output vector that is color corrected.

An image processing device and method are provided for adjusting background pixels of an image. The device includes memory which stores a background adjustment component which for each of a plurality of pixels of an input image: computes a background strength of the pixel; computes a luminance strength of the pixel; and computes adjusted luminance and adjusted chrominance values for the pixel, as a function of the background strength and luminance strength of the pixel. An image output component outputs an output image derived from the adjusted luminance and adjusted chrominance values for the plurality of pixels, A processor implements the background adjustment component and image output component.

Techniques related to removing haze from video are discussed. Such techniques include color converting a video frame from an input color space to a haze color space using a haze color detected in a previous frame, estimating pixel-wise haze amounts, de-hazing the video frame in the haze color space using the pixel-wise haze amounts, and color converting the de-hazed frame to the input color space.

A method, apparatus, and non-transitory computer readable medium for color reduction based on image segmentation are described. The method, apparatus, and non-transitory computer readable medium may provide for segmenting an input image into a plurality of regions, assigning a weight to each region, identifying one or more colors for each of the regions, selecting a color palette based on the one or more colors for each of the regions and the corresponding weight for each of the regions, and performing a color reduction on the input image using the selected color palette to produce a color reduced image. The weight assigned to each region may depend on factors including relevance, prominence, focus, position, or any combination thereof.

System that makes nutritional recommendations based on results of a home urine test. A user may apply a urine sample to a card containing multiple tests, and capture an image of the card using a phone; an analysis system executing on the phone or in the cloud may analyze the image and determine test results. The test card and analysis system may compensate for variability in lighting conditions and time of exposure to the urine sample. Based on test results, the system may recommend consumption of specific quantities of nutrients, such as vitamins, minerals, and foods. It may also recommend consumption of water or electrolytes based on measured hydration, and stress reduction techniques or sleep based on measured cortisol. Recommendations may be customized based on factors such as the user's characteristics (gender, weight, etc.), predicted absorption of nutrients from food or supplements, and the user's dietary preferences or restrictions.

Example method includes: transmitting a plurality of probe images from an IoT device at an edge network to a server hosting a target DNN, wherein the plurality of images are injected with a limited amount of noise to probe sensitivities of the target DNN to the red, green, and blue colors; receiving a feedback comprising a plurality of DCT coefficients unique to target DNN from the server hosting the target DNN; computing a plurality of color conversion weights based on the feedback received from the server; converting a set of real-time images from RGB color space to YUV color space using the plurality of color conversion weights unique to the target DNN; compressing the set of real-time images using a quantization table unique to the target DNN by the IoT device; and transmitting the compressed set of real-time images to the server hosting the target DNN for DNN inferences.

An image processing apparatus performs predetermined image processing for read data obtained by reading an image on a document by first and second reading sensors arranged so as to have an overlapping area. The predetermined image processing includes color correction processing of converting color signals detected by light receiving elements of the reading sensors into color signals with respect to each of pixels forming read data. Color correction processing for first read data obtained from light receiving elements of the first reading sensor not included in the overlapping area and color correction processing for second read data obtained from light receiving elements of the second reading sensor not included in the overlapping area are performed in accordance with different correction characteristics. By doing this, an image in which image unevenness caused by individual differences of the short reading sensors is reduced can be output.

A method for processing an image may include: collecting an ambient light spectrum while capturing an image (10); determining RGB tristimulus values corresponding to ambient environment of the captured image based on the collected ambient light spectrum and a color filter absorption curve (11); determining skin color coordinates based on the collected ambient light spectrum and a spectral absorption curve of a skin color model (12); and determining RGB values in the skin color model based on the RGB tristimulus values and the skin color coordinates (13).