A New Feedback-Based Method for Parameter Adaptation in Image Processing Routines

Abstract

The parametrization of automatic image processing routines is time-consuming if a lot of image processing parameters are involved. An expert can tune parameters sequentially to get desired results. This may not be productive for applications with difficult image analysis tasks, e.g. when high noise and shading levels in an image are present or images vary in their characteristics due to different acquisition conditions. Parameters are required to be tuned simultaneously. We propose a framework to improve standard image segmentation methods by using feedback-based automatic parameter adaptation. Moreover, we compare algorithms by implementing them in a feedforward fashion and then adapting their parameters. This comparison is proposed to be evaluated by a benchmark data set that contains challenging image distortions in an increasing fashion. This promptly enables us to compare different standard image segmentation algorithms in a feedback vs. feedforward implementation by evaluating their segmentation quality and robustness. We also propose an efficient way of performing automatic image analysis when only abstract ground truth is present. Such a framework evaluates robustness of different image processing pipelines using a graded data set. This is useful for both end-users and experts.

AutoCellSeg: robust automatic colony forming unit (CFU)/cell analysis using adaptive image segmentation and easy-to-use post-editing techniques

“…A BLOB plausibility check is performed over the resulting image where any b i having size a i < a min ( a min is derived from fuzzy a priori knowledge as described in Section 5) is removed. Furthermore, bigger BLOBs with a i > a big (where a big is the upper 30% of median a i ) are passed onto feedback-based watershed segmentation step which adapts the value of H-maxima transform for regional maxima of I in 26 . The regional maxima are used as seed points for watershed segmentation in the automatic mode of AutoCellSeg (see Section 6).…”

“…a min = min ( a a | i ) and a max = max ( a a | i )) are calculated for both of them. Thereafter, a fuzzy trapezoidal membership function μ ( a ) (according to 26 ) is used. The edges of trapezoid are ( p 1 , p 2 , p 3 , p 4 ) = (0.5 a min , a min , a max , 2 a max .…”

“…This method has been adapted from 26 , where it was applied to adapt intensity threshold for segmentation. The watershed segmentation 27 is quite useful in separating chunks of BLOBs.…”

Axial and radial development of the hot electron distribution in a helicon plasma source, measured by a retarding field energy analyzer (RFEA)

“…We further apply the watershed algorithm following Khan et al (2016) and Khan et al (2018), which we modify and expand to handle colony confluency. Here, distance transformation along multi-threshold-based watershed is consolidated with quality Watershed segmentation relies on a topographic (intensity) information across two spatial coordinates, x and y, reflecting the colony number in each BLOB.…”

Computer Vision and Image Processing: A Paper Review

“…For example, in face processing, the algorithm is used to identify the element of head and face which has a lower resolution [49], [50]. The algorithm also improves the machine performance to determine the eyes, eyebrows, and mouth [51], [52], [53]. Whereas, the final parts such as ears and neck are rarely studied in face processing.…”