Abstract: Systems and methods for sonar image feature extraction, which fit a parameterized statistical model to sonar image data to extract features that describe image textures for follow-on pattern classification tasks. The systems and methods estimate the parameters of an assumed statistical model from the pixel values in the sonar images. The parameters of the distribution can then be used as features in a discriminant classifier to perform tasks such as texture segmentation or environmental characterization. The systems and methods divide a sonar image into separate overlapping cells. The ACF parameters of each cell are sequentially estimated using various methods. Based on the ACF parameters obtained, each cell is assigned a label via a pattern classification algorithm. The labeled image can be used in target recognition, environmental characterization and texture segmentation.

Abstract: Line contrast difference effect which limits the capabilities of laser line scan (LLS) sensors is substantially reduced by a process that uses a smhing array, modified based upon a mean magnitude array of all columns of a discrete Fourier transform (DFT) of an image matrix. Trouble frequencies (both small band width and large band width) indicative of line contrast difference effect are identified from the mean magnitude DFT array using a linear least squares error method. The same technique is used to calculate suppression values for each of the identifiable trouble frequencies and the smoothing array is modified thereby. Once all modifications are made to the smoothing array, it is applied to the mean magnitude DFT array to suppress those pixels of the array that are associated with trouble frequencies. The corrected mean magnitude DFT array is then subject to an inverse DFT function for each column in the array to reconstruct the original image, now in corrected form, to be saved for display or storage.

Abstract: A method for detecting a target partially obscured in the image. Sensor data from an area of concern in converted to image data indexed to represent the area. Test signature data is provided representative of the image of a portion of the type of target of interest. The image data is normalized and filtered using the test signature data to increase the signal-to-noise ratio of portions of the normalized data that approximate the test signature data. The so-filtered data is digitized to a logic level "1" or "0" based upon threshold criteria. The resulting digitized data is convolved with a unity weighting function window to generate correspondingly indexed density data. The density data values whose first conditional probability is at least p times more likely than the second conditional probability are then converted to a logic level "1" while the density data values whose first conditional probability is less than p times more likely than the second conditional probability are converted to a logic level "0".