HydroCharting is proud to present these fine Seabed Classification tools to our Scandinavian customers:

• Use your analogue echosounder with QTC5.5 and QTRT for real-time seabed classification, while, if desired, logging the raw data to post process in QTC IMPACT.

• Directly post-process your data from a supported fully digital echosounder in QTC IMPACT, and, if desired, add QTRT for real-time classification.

Classification in water less than about 5 meters deep presents unique challenges. Quester Tangent recognized this limitation and developed a high-speed data acquisition system for shallow water echoes, the QTC5.5. In such shallow water, four issues must be addressed.

• The sampling rate must be fast enough to capture the details in very short echoes. Shallow water echoes are always brief, as there is little time for the sound to spread across the seabed.

• Triggering must be very reproducible because the transmit pulse and the echoes can be barely separated and classification requires that all but the signal from the seabed is excluded.

• Data acquisition systems are usually designed to capture echoes from depths of several hundred meters. The amplitudes of echoes from less than a meter can be much larger, so the acquisition system needs a remarkable dynamic range.

• Echoes from identical sediments at different depths are different, particularly in duration, and so are features from those echoes. In deep water, compensation for depth changes can be linear, but this is not true in shallow water.

Even if the bottom in all or part of the survey area could be classified visually, which is not always possible or practical, acoustic bottom classification may well be preferred so that the survey is systematic and consistent over a large area. In fact, the opportunity to classify visually can assist the acoustic results by providing ground truth over large areas for comparison with the acoustic results, rather than the usual sparse set of point samples. At the other end of the scale recent work with the QTC5.5 explored classification in greater water depths. Good data has successfully been collected in waters as deep as 2260 meters.

The QTC5.5 acquisition hardware samples at up to 1.25 MHz, and then uses Digital Signal Processing to decimate to the right rate for the particular water depth. No other system samples faster than 44 kHz. By fully exploiting the capabilities of DSP, the QTC5.5 produces echo envelopes which are considerably more accurate than those generated by analogue rectification. QTC5.5 is integrated with QTRT acquisition software and QTC IMPACT post-processing software. Together they are a unique system with superior classification performance that is not available anywhere else.

QTC5.5 follows a simple philosophy: connect the system to a regular single beam echosounder, capture the fundamental data by recording a high fidelity version of the echo waveform and apply the proven Quester Tangent process to generate a map of bottom type. Using QTRT acquisition software, raw echo information can be collected for post-processing, while classification results can be simultaneously overlaid and displayed in real-time on a seabed chart.

The QTC5.5 can be supplied as a turnkey system that includes echosounder, QTC5.5 Sounder Interface Module, GPS and customized computer with all necessary software, or individual components can be supplied as required to match a client’s existing hardware. The system can be used for simple repetitive surveying or scaled up to become a sophisticated research tool providing a wealth if information about the nature of a sea, lake or river bed.

QTC IMPACT

QTC IMPACT is software that processes digital single beam data (echoes) to make maps of bottom type. The software permits the user to process data without the need to build a catalogue or database of seabed types prior to data processing. Echo data are processed into discrete acoustic classes using a feature generation and 3-D clustering routine. The result is a logical grouping of the data into representative acoustic classes which can then be plotted on a map of the seabed with the boundaries of the classes clearly defined, i.e. the user builds a map of acoustic diversity of the seabed of the surveyed area. Echo-shape measurement and feature generation is done using the same processing techniques and algorithms as used in QTC5.5 and QTRT.

The resulting information (set of classified echoes) can be correlated with other data types the user may have on hand, such as coring data, fish catch statistics, video data, local knowledge etc. Alternatively, bottom type can be related to acoustic class data by directed physical ground truthing.

The classification result can also be used to define a catalogue for classifying future or archived data sets, or for real time survey with a QTC5.5. This speeds up future processing times, as one would simply log the data and then run it through the catalogue already pre-defined from earlier work to provide a classification of the area surveyed. The procedure using a pre-defined catalogue of classes is referred to as “Supervised Classification”, while “Unsupervised Classification” generally relates to classifying a data set without the use of any pre-defined catalogue or classification scheme.

As processing in QTC IMPACT is done in post-processing rather than real-time, this enables more control in respect of data manipulation and quality assurance. Data filtering permits the removal of any bad or unwanted data points to prevent introducing artifacts in the classification, and the clustering process can be used to provide an optimum level of classification. For instance a dataset collected from a geologically diverse area, e.g. ranging from rock to mud, when submitted to QTC IMPACT for initial classification, is not likely to directly provide classification between different sand types (for example) found within the area. However, by isolating the sand types as a subset of data, and then submitting only this subset to QTC IMPACT for processing, the system will then be able to detect more subtle differences in the acoustic signatures from the sand.

QTC IMPACT will directly load data from a number of sources. These currently include QTC VIEW data and data from fully digital echosounders such as manufactured by Biosonics, Knudsen, Odom and Simrad. QTC IMPACT outputs data in a comma delimited user-customizable ASCII file, the “.seabed” file, ready for direct import into GIS or mapping software, or for added-value processing in QTC CLAMS.

QTC SWATHVIEW is software that provides classification of multibeam, sidescan or interferometric swath sonar image data. There is no hardware involved – it simply uses the backscatter data as logged by the swath sonar. Whereas single beam classification is based on the shape of the echo (i.e. time-series classification), swath classification is based on the intensity and texture of the backscatter image. QTC SWATHVIEW will load and process data from almost all the major multibeam and sidescan systems in operation today.


QTC SWATHVIEW can be summarized as follows:

• Software that accepts image data collected with a swath sonar system that records backscatter intensities. The software operates on a Windows platform.

• Deploys a large number of rectangular patches on the image, in positions appropriate to each type of sonar.

• Applies a suite of algorithms to generate many statistical features describing the backscatter and depth (where appropriate) of each patch of seabed.

• Applies processing to generate feature combinations that best capture the diversity of the data.

• Classifies the feature combinations to produce statistically valid clusters representing seabed types.

• The software accepts backscatter intensities and navigational/attitude information in the format appropriate for the sonar system.

• Raw line-by-line image data are used. No mosaic imagery is used for classification, as the mosaic is sub-sampled data, and contains artefacts that will affect the result.

• Quality control is applied by disregarding data associated with bad bottom picks, depression angles out of range (compensation for angle of incidence), or insufficient positioning data. Details depend on the specific sonar situation. Good quality control is very important for good results.

• Compensation for the effects of sonar or grazing angle on backscatter is applied. The backscatter reflection is stronger under the ship versus weaker at an angle away from the ship, from the same seabed type, and this must be adjusted (compensated) to the same levels. This process is patented and is one of the key features of the software.

• Rectangles (patches) are deployed across the image in locations appropriate to the type of sonar system. The information in these patches is used by a suite of algorithms to generate features to describe the image texture and intensity. These features, known as Full Feature Vectors (FFV), are then used to segment the image into discrete classes.

• As in QTC IMPACT, the software utilizes a catalogue (also referred to as a library) that can be updated or edited. QTC SWATHVIEW allows both supervised and unsupervised classification of a data set. Supervised classification is based on an existing library. Unsupervised classification allows the catalogue to be developed as part of the classification process.

• The data output consists of acoustic classes for all seabed patches, in a user-customizable comma-delimited ASCII format for use by mapping software or a GIS. The standard data output file has an identical format to the *.seabed single beam final output file.

QTC CLAMS

CLAMS is a software utility package that provides for data enhancement, visualization and presentation. It is a collection of utilities that can be used to manipulate categorical, or class-type, data. Data types that can be read by QTC CLAMS include:

• QTC data formats, e.g. .seabed and .dat files

• other acoustic class data such as RoxAnn or ECHOplus E1 and E2 data

• non-acoustic xyz data such as video class data or grab sample class data.

QTC CLAMS provides the following processing options:

• Categorical Interpolation: an extensive investigation and search of existing products and technology was unable to find any interpolation routines suitable for categorical data arranged in discrete classes or categories. For example, in discrete class data there may be two classes that are spatially separated, the first being class 1, the second being class 3. This does not necessarily mean that class 2 is between them. The regular interpolation routines currently available all use linear functions, and would place class 2 between classes 1 and 3. QTC CLAMS intelligently "fills in the gaps" between discrete classes.

• Similarity Colours: this algorithm assigns colours to classes based on their acoustic diversity. Acoustic classes that are similar to each other will be painted with similar colours, rather than the traditional approach of assigning discrete colours. This provides for a much-improved visualisation of a classified map as the display colours have some meaning.

• Complexity and Variability: these are two different ways to measure the nature of change of a seabed over a given area. This process uses the variance of the Q-values and distribution of class values to determine how complex or variable a seabed is.