Real-time filtering on parallel SIMD architectures for automated quality inspection

Usamentiaga, Rubén

doi:10.1007/s11554-020-00954-3

Cited by 2 publications

(5 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Comparing the results with previous works, for example in [31], the proposed procedure is around 4 times faster on similar hardware. These results confirm that the proposed approach to transform filtering into GEMM operations provides better results than directly designing ad hoc procedures based on SIMD and multicore parallelism.…”

Section: The Double Fifo Filter and The Double Circular Filtersupporting

confidence: 50%

“…An optimized implementation for FIR filters on multicore CPUs is presented in [31]. This work proposes different strategies to filter height maps in real-time.…”

Section: Filtering In Surface Metrologymentioning

confidence: 99%

“…In previous works, the implementation of the filtering techniques was optimized for particular CPU and GPU architectures based on specific parallel execution models. For example, in [31] the optimization is based on parallels execution on multicore CPUs taking advantage of SIMD instructions (single instruction, multiple data). In [28] specific CUDA kernels are implemented to run the filtering process on multiple GPUs.…”

Section: Real-time Filtering Approachmentioning

confidence: 99%

“…After the matrix is filled, a steady state is reached where the most recent profiles are always stored in a contiguous block of memory starting from the last insert position. Further details are given in [31].…”

Section: The Double Fifo Filter and The Double Circular Filtermentioning

confidence: 99%

“…Following this approach, at any moment the most recent profiles are always stored in a contiguous block of memory starting from the top index. Further details are given in [31].…”

Section: The Double Fifo Filter and The Double Circular Filtermentioning

confidence: 99%

See 4 more Smart Citations

CPU and GPU real-time filtering methods for dense surface metrology using general matrix to matrix multiplications

Usamentiaga

2022

J Real-Time Image Proc

Self Cite

View full text Add to dashboard Cite

Filtering is a required task in surface metrology for the identification of the components relevant for automated quality control. The calculation of real-time features about the surface is crucial to determining the mechanical and physical properties of the inspected product. The computation efficiency of the filtering operations is a major challenge in surface metrology, as current sensors provide massive volumes of data at very high acquisition rates. To overcome the challenges, this work presents different real-time filtering solutions comparing the performance on the CPU and on the GPU, using modern hardware. The proposed framework is focused on filtering techniques that can be expressed using a finite impulse response (FIR) kernel that includes the Gaussian kernel, the most common filtering technique recommended by ISO and ASME standards. This research work proposes variations of the double FIFO and double circular filters. The filters are transformed into a series of general matrix to matrix multiplications, which can be run extremely efficiently on different architectures. The proposed filtering approach provides superior performance compared with previous works. Additionally, tests are carried out to quantify the performance of the GPU in terms of data transfer and computation capabilities in order to diminish the penalty imposed by data transfer from main memory to the GPU in real-time operations. Based on the results, an efficient batch filtering technique is proposed that can be run on the GPU faster than the CPU even for small profile and kernel sizes, offloading this task from the host CPU for optimal system and application response.

show abstract

Section: The Double Fifo Filter and The Double Circular Filtersupporting

confidence: 50%

“…An optimized implementation for FIR filters on multicore CPUs is presented in [31]. This work proposes different strategies to filter height maps in real-time.…”

Section: Filtering In Surface Metrologymentioning

confidence: 99%

Section: Real-time Filtering Approachmentioning

confidence: 99%

Section: The Double Fifo Filter and The Double Circular Filtermentioning

confidence: 99%

“…Following this approach, at any moment the most recent profiles are always stored in a contiguous block of memory starting from the top index. Further details are given in [31].…”

Section: The Double Fifo Filter and The Double Circular Filtermentioning

confidence: 99%

See 3 more Smart Citations