The author presents a technique to calculate instrument stellar magnitudes for silicon photosensors used in modern star trackers as well as estimates the error of the said technique. The technique implies calculation of instrument stellar magnitudes as specifically selected functions of color indices B-V, V-J of the Tycho-2 and two micron all sky survey (2MASS) catalogs. This function is a sum of basis functions with coefficients determined based on star trackers' response. The coefficients are calculated individually for each star tracker response. Calculation of the coefficients of the function is done using the least squares method for color indices of artificial stars. Spectra of artificial stars are created on the basis of reddening of typical subclasses spectra without interstellar extinction, such spectra taken from the Pickles catalog. Thickness of interstellar medium for reddening is selected basing on random law. Validation of accuracy of the proposed methodology is performed by calculating star brightness for standard photometric bands R and I and subsequent comparison of the obtained results with actual data. Such check indicates that the root-mean-square deviation of error is not over 0.08 m. Due to usage of data related to star brightness in J band of 2MASS catalog, the number of stars in the guide star catalog is increased by 30% as compared with using data only from Tycho and HIPPARCOS catalogs.

Layout estimation is a challenge of segmenting a cluttered room image into floor, walls and ceiling. We applied Double refinement network proved to be efficient in the depth estimation to generate heat maps for room key points and edges. Our method is the first not using encoder-decoder architecture for the room layout estimation. ResNet50 was utilized as a backbone for the network instead of VGG16 commonly used for the task, allowing the network to be more compact and faster. We designed a special layout score function and layout ranking algorithm for key points and edges output. Our method achieved the lowest pixel and corner errors on the LSUN data set. The input image resolution is 224*224.

No abstract

No abstract

No abstract