Abstract: Methods and apparatuses for correcting bad image pixels are described. The described sensor-independent image processing techniques leverage one or more dynamic dictionaries of learned filters for bad pixel correction (e.g., where a camera leverages such dictionaries to efficiently identify filters to accurately adjust and correct bad pixel values). For example, a dictionary may store filters that are learned offline (via a self-supervised learning algorithm implemented at a server using known images and ground truth bad pixel correction values). To select a filter for a bad pixel correction operation, a camera may encode an image patch surrounding a bad pixel (into an encoded patch descriptor) and search the dictionary for a matching patch descriptor key. The camera may then apply the filter (value) corresponding to the searched patch descriptor (key) of the dictionary to the image patch to correct the bad pixel and generate a corrected output image.

Abstract: In a method for time-of-flight (ToF) based measurement, a scene is illuminated using a ToF light source modulated at a first modulation frequency FMOD(1). While the light is modulated using FMOD(1), depths are measured to respective surface points within the scene, where the surface points are represented by a plurality of respective pixels. At least one statistical distribution parameter is computed for the depths. A second modulation frequency FMOD(2) higher than FMOD(1) is determined based on the at least one statistical distribution parameter. The depths are then re-measured using FMOD(2) to achieve a higher depth accuracy.

Abstract: Methods and apparatuses for correcting bad image pixels are described. The described sensor-independent image processing techniques leverage one or more dynamic dictionaries of learned filters for bad pixel correction (e.g., where a camera leverages such dictionaries to efficiently identify filters to accurately adjust and correct bad pixel values). For example, a dictionary may store filters that are learned offline (via a self-supervised learning algorithm implemented at a server using known images and ground truth bad pixel correction values). To select a filter for a bad pixel correction operation, a camera may encode an image patch surrounding a bad pixel (into an encoded patch descriptor) and search the dictionary for a matching patch descriptor key. The camera may then apply the filter (value) corresponding to the searched patch descriptor (key) of the dictionary to the image patch to correct the bad pixel and generate a corrected output image.

Abstract: In a method for time-of-flight (ToF) based measurement, a scene is illuminated using a ToF light source modulated at a first modulation frequency FMOD(1). While the light is modulated using FMOD(1), depths are measured to respective surface points within the scene, where the surface points are represented by a plurality of respective pixels. At least one statistical distribution parameter is computed for the depths. A second modulation frequency FMOD(2) higher than FMOD(1) is determined based on the at least one statistical distribution parameter. The depths are then re-measured using FMOD(2) to achieve a higher depth accuracy.

Abstract: In a method for time-of-flight (ToF) based measurement, a scene is illuminated using a ToF light source modulated at a first modulation frequency FMOD(1). While the light is modulated using FMOD(1), depths are measured to respective surface points within the scene, where the surface points are represented by a plurality of respective pixels. At least one statistical distribution parameter is computed for the depths. A second modulation frequency FMOD(2) higher than FMOD(1) is determined based on the at least one statistical distribution parameter. The depths are then re-measured using FMOD(2) to achieve a higher depth accuracy.

Abstract: A method for time-of-flight (ToF) guided down-up sampling includes calculating a guide function from a full resolution output of a ToF sensor, downsampling the full resolution output by a predetermined scaling factor, calculating a downsampled depth data from the downsampled output and upsampling the downsampled depth data to full resolution using the full resolution guide function.

Abstract: A method for time-of-flight (ToF) guided down-up sampling includes calculating a guide function from a full resolution output of a ToF sensor, downsampling the full resolution output by a predetermined scaling factor, calculating a downsampled depth data from the downsampled output and upsampling the downsampled depth data to full resolution using the full resolution guide function.

Abstract: In a method for time-of-flight (ToF) based measurement, a scene is illuminated using a ToF light source modulated at a first modulation frequency FMOD(1). While the light is modulated using FMOD(1), depths are measured to respective surface points within the scene, where the surface points are represented by a plurality of respective pixels. At least one statistical distribution parameter is computed for the depths. A second modulation frequency FMOD(2) higher than FMOD(1) is determined based on the at least one statistical distribution parameter. The depths are then re-measured using FMOD(2) to achieve a higher depth accuracy.

Abstract: In a method for time-of-flight (ToF) based measurement, a scene is illuminated using a ToF light source modulated at a first modulation frequency FMOD(1). While the light is modulated using FMOD(1), depths are measured to respective surface points within the scene, where the surface points are represented by a plurality of respective pixels. At least one statistical distribution parameter is computed for the depths. A second modulation frequency FMOD(2) higher than FMOD(1) is determined based on the at least one statistical distribution parameter. The depths are then re-measured using FMOD(2) to achieve a higher depth accuracy.

Abstract: In a method for time-of-flight (ToF) based measurement, a scene is illuminated using a ToF light source modulated at a first modulation frequency FMOD(1). While the light is modulated using FMOD(1), depths are measured to respective surface points within the scene, where the surface points are represented by a plurality of respective pixels. At least one statistical distribution parameter is computed for the depths. A second modulation frequency FMOD(2) higher than FMOD(1) is determined based on the at least one statistical distribution parameter. The depths are then re-measured using FMOD(2) to achieve a higher depth accuracy.

Abstract: A method, and one or more non-transitory computer-readable media storing instructions, and a device for removal of unwanted components in an audio signal, the device comprising a processor, coupled to memory, configured to receive reference and processed inputs into memory where the processed input is a result of a reduction process of unwanted components of the audio signal, estimate envelope values for processed and reference inputs at a plurality of time and frequency instances, for each time and frequency instance: compute a first gain in relation to a ratio of the estimated envelope value of the processed input to the estimated envelope value of the reference input, apply a nonlinear process to said first gain to produce a second gain, compute an output gain as the ratio between second gain and first gain and, apply the output gain to processed input, thereby producing a filtered output with unwanted components suppressed.

Abstract: A device for removal of unwanted components in an audio signal, the device comprising a processor, coupled to memory, configured to receive reference and processed inputs into memory where the processed input is a result of a reduction process of unwanted components of the audio signal, estimate envelope values for processed and reference inputs at a plurality of time and frequency instances, for each time and frequency instance: compute a first gain in relation to a ratio of the estimated envelope value of the processed input to the estimated envelope value of the reference input, apply a nonlinear process to said first gain to produce a second gain, compute an output gain as the ratio between second gain and first gain and, apply the output gain to processed input, thereby producing a filtered output with unwanted components suppressed.